Exhibit 99.1

Technical Report on the Kibali Gold Mine, Democratic Republic of the Congo

Report for NI 43-101

Randgold Resources Limited

3^rd Floor Unity Chambers 28 Halkett Street St Helier Jersey OJE2

18^th September 2018

Effective Date: 31^st December 2017

Qualified Persons:

Mr. Rodney B. Quick, MSc, Pr. Sci.Nat

Mr. Simon Bottoms, CGeol, MGeol, FGS, MAusIMM

Mr. Richard Quarmby, BSc, Pr Eng, C Eng, MSAIChE, MIoMMM, MBA

Mr. Andrew Law, HND(MMin), MBA, FAusIMM (CP), FIQA, MAICD, AFAIM

Mr. Graham E. Trusler, Msc, Pr Eng, MIChE, MSAIChE

Kibali Gold Mine NI 43-101 Technical Report

Table of Contents
1   Executive Summary		1
1.1	Property Description and Location	2
1.2	Mineral Rights and Land Ownership	2
1.3	History	4
1.4	Geology and Mineralisation	5
1.5	Exploration	8
1.6	Mineral Resources	8
1.7	Ore Reserves	10
1.8	Mining Method	12
1.9	Mineral Processing	14
1.10	Project Infrastructure	18
1.11	Market Studies	20
1.12	Environmental, Permitting and Social Considerations	21
1.13	Capital Costs	22
1.14	Operating Costs	23
1.15	Economic Analysis	24
1.16	Interpretation and Conclusions	25
1.17	Risks	28
1.18	Recommendations	31
2   Introduction		32
2.1	Introduction	32
2.2	Effective Date	33
2.3	Sources of Information	33
2.4	List of Abbreviations	34
2.5	Units	36
3   Reliance on Other Experts		37
4   Property Description and Location		38
4.1	Project Location	38
4.2	Mineral Rights and Land Ownership	38
4.3	Surface Rights	42
4.4	Ownership, Royalties and Lease Obligations	42
5   Accessibility, Climate, Local Resources, Infrastructure and Physiography		45
5.1	Accessibility	45
5.2	Climate	45
5.3	Local Resources	46

18^th September 2018 Page ii

Kibali Gold Mine NI 43-101 Technical Report

5.4	Infrastructure	47
5.5	Physiography	49
6   History		51
6.1	Historical Exploration and Development	51
6.2	Kibali Project Milestones and Development	52
6.3	Historical Resource and Reserve Estimates	53
6.4	Past Production	55
7   Geological Setting and Mineralisation		56
7.1	Regional Geology	56
7.2	Structural Geology	58
7.3	Local Geology	59
7.4	Mineralisation	62
7.5	Project Deposits	64
8   Deposit Types		71
9   Exploration		72
9.1	Exploration Concept	72
9.2	Historical Exploration Review	72
9.3	Geophysics	73
9.4	Geochemical Sampling	76
9.5	Proposed 2018 Regional Exploration	77
9.6	Proposed 2018 Resource Definition Exploration	77
9.7	Discussion	77
10   Drilling		78
10.1	Drill Hole Database	78
10.2	Survey Grid	79
10.3	Drill Planning and Site Preparation	79
10.4	Downhole Surveying	79
10.5	Collar Surveys	79
10.6	Diamond Drilling	80
10.7	Reverse Circulation Drilling	81
10.8	Drill Twinning Studies	81
10.9	Kibali Mineral Resource Drill Spacing Optimisation	82
10.10	Other Sampling Methods	82
10.11	Discussion	83
11   Sample Preparation, Analyses and Security		84
11.1	Sample Selection	84
11.2	Sample Preparation	84

18^th September 2018 Page iii

Kibali Gold Mine NI 43-101 Technical Report

11.3	Sample Analysis	88
11.4	Quality Assurance and Quality Control	88
11.5	Security	110
11.6	Independent Audits	111
11.7	Discussion	111
12   Data Verification		114
12.1	Historical Drill Hole Data Verification	114
12.2	Kibali Drill Hole Data Verification	114
12.3	Independent Audit	115
13   Mineral Processing and Metallurgical Testing		116
13.1	Summary	116
13.2	Testwork Strategy & Sample Selection: Extraction	118
13.3	Open Pit Operations	127
13.4	Metallurgical Recoveries	132
13.5	Deleterious Elements	132
14   Mineral Resource Estimates		134
14.1	Summary	134
14.2	Resource Database	137
14.3	Geological Modelling	146
14.4	Topography	156
14.5	Bulk Density	156
14.6	Compositing	162
14.7	Treatment of High-grades (Top Cutting)	167
14.8	Variography	171
14.9	Block Model Estimation	181
14.10	Block Models	185
14.11	Resource Classification	191
14.12	Block Model Depletion	192
14.13	Block Model Validation	193
14.14	Resource Cut-Off Grades	195
14.15	Mineral Resources Reporting	207
14.16	2017 Versus 2016 Mineral Resource Comparison	210
14.17	Reconciliation	215
14.18	Discussion	217
15   Ore Reserve Estimate		221
15.1	Summary	221
15.2	Ore Reserve Estimation Process	222

18^th September 2018 Page iv

Kibali Gold Mine NI 43-101 Technical Report

15.3	Geotechnical and Hydrogeological Considerations	229
15.4	Dilution and Mining Recovery	242
15.5	Economic Parameters	244
15.6	Open Pit Optimisations	248
15.7	Mine Design	258
15.8	External Audits	264
16   Mining Methods		265
16.1	Open Pit Method	265
16.2	Underground Method	269
16.3	Life-of-Mine Plan	272
16.4	Underground Mine Infrastructure and Services	281
17   Recovery Methods		283
17.1	Processing Plant	283
17.2	Processing Recovery	292
17.3	Production History	295
17.4	Capital Projects and Plant Upgrade	296
17.5	Processing Costs	298
18   Project Infrastructure		300
18.1	Mine Roads	300
18.2	Supply Chain	300
18.3	Surface Water Management	301
18.4	Water Supply	302
18.5	Tailings Facilities	302
18.6	Power Supply	304
18.7	Site Infrastructure	306
18.8	Communication and Information Technology	308
18.9	Security	308
19   Market Studies and Contracts		309
19.1	Revenue, Tax and Royalty	309
19.2	Marketing	309
19.3	Contracts	309
20   Environmental Studies, Permitting, And Social or Community Impact		311
20.1	Environmental Considerations	311
20.2	Social Considerations	318
20.3	Artisanal and Small-Scale Mining	322
21   Capital and Operating Costs		324
21.1	Capital Costs	324

18^th September 2018 Page v

Kibali Gold Mine NI 43-101 Technical Report

21.2	Operating Costs	325
22   Economic Analysis		327
23   Adjacent Properties		328
23.1	Kibali South	328
23.2	Zani Kodo	328
24   Other Relevant Data and Information		329
24.1	Country Risk	329
24.2	DRC Mining Code Review	330
25   Interpretation and Conclusions		332
25.1	Geology and Mineral Resources	332
25.2	Mining and Ore Reserves	332
25.3	Processing	333
25.4	Environment and Social	334
25.5	Ownership and DRC Mining Code	335
25.6	Infrastructure	335
25.7	Risks	336
26   Recommendations		339
27   References		340
28   Date and Signature Page		344
29   Certificate of Qualified Persons		345
29.1	Simon P. Bottoms, CGeol, MGeol, FGS, MAusIMM	345
29.2	Rodney B. Quick, MSc, Pr. Sci.Nat	346
29.3	Richard Quarmby, Pr Eng, C Eng, SAIChE	347
29.4	Andrew Law, HND(MMin), MBA, FAusIMM (CP), FIQA, MAICD, AFAIM	348
29.5	Graham E. Trusler, Pr Eng, IChE	349
30   Appendix		350
30.1	Appendix 1 – JORC 2012 Edition – Table 1	350

18^th September 2018 Page vi

Kibali Gold Mine NI 43-101 Technical Report

List of Tables
Table 1-1 Kibali Exploitation Permit Details	3
Table 1-2 Mineral Resource Statement for the Kibali Mine as of 31st December 2017	9
Table 1-3 Kibali Mine Ore Reserve Summary as of 31st December 2017	11
Table 1-4 Summary of Average Recovery for All the Samples	16
Table 1-5 Actual Process and Plant Engineering Operating Costs for 2016 and 2017	18
Table 1-6 LOM Capital Expenditure	23
Table 1-7 LOM Unit Operating Costs	24
Table 1-8 Kibali Risk Analysis	30
Table 4-1 Kibali Exploitation Permit Details	40
Table 4-2 Kibali Exploitation Permit Coordinates	41
Table 4-3 Kibali Exploitation Permit Coordinates	42
Table 6-1 Summary of Historical Kibali Trenches, Auger and Pits Summary	52
Table 6-2 Moto Goldmines Ltd. Mineral Resource Estimate as of August 2008	54
Table 6-3 Moto Goldmines Ltd. Ore Reserve Estimate as of August 2008	54
Table 6-4 Past Production Records for the Kibali Mine	55
Table 9-1 Kibali Trenches, Auger and Pits Summary	76
Table 10-1 Kibali Drilling Summary	78
Table 11-1 2017 Submitted Samples	88
Table 11-2 List of CRMs Assayed from SGS Doko	92
Table 11-3 CRM Summary for Review Period at SGS Doko	92
Table 11-4 CRM Statistics for SGS Doko	92
Table 11-5 List of CRMs Assayed at SGS Mwanza	95
Table 11-6 CRM Summary for SGS Mwanza	95
Table 11-7 CRM Statistics for SGS Mwanza	95
Table 11-8 Statistics for Blank Samples at SGS Doko	98
Table 11-9 Statistics for Blank Samples at SGS Mwanza	99
Table 11-10 Statistics for RC Duplicates at SGS Doko	100
Table 11-11 Statistics of Coarse Duplicates at SGS Doko	102
Table 11-12 Statistics for Coarse Duplicates at SGS Mwanza	105
Table 11-13 Statistics of Trench Duplicates at SGS Doko	107
Table 11-14 Summary of Pulp Duplicates Analysed at OMAC	110
Table 13-1 Summary of Testwork	116
Table 13-2 Physical and Extraction Sample Selection and Testwork Logic (Moto FS_2007- 1377\16.14\1377-STY-002\S5)	120
Table 13-3 Extraction Comparison - Underground Variability	122
Table 13-4 Isolated Samples for Further Analysis	123

18^th September 2018 Page vii

Kibali Gold Mine NI 43-101 Technical Report

Table 13-5 Direct Cyanidation Results	127
Table 13-6 Metallurgical Recoveries Used for Deposits in Pit Optimisation	129
Table 13-7 KCD Fresh Open Pit Fresh Samples - Lode 5000	131
Table 13-8 Summary of Average Recovery for All the Samples	132
Table 14-1 Kibali Mine Mineral Resource Statement as of 31st December 2017	134
Table 14-2 Summary of Deposits and Model Date	135
Table 14-3 Drill Summary of KCD Used in 2017 Mineral Resource Estimate	137
Table 14-4 KCD Open Pit Resources Composite Data – 2017 Mineral Resource Estimate	138
Table 14-5 KCD 3000 and 5000 Underground Resources Composite Data – 2017 Mineral Resource Estimate	138
Table 14-6 KCD 9000 Underground Mineral Resource Estimates Composite Data – 2017 Mineral Resource Estimate	139
Table 14-7 Sessenge Drill Summary of Holes Used in 2017 Mineral Resource Estimate	139
Table 14-8 Sessenge Composite Data – 2017 Mineral Resource Estimate	140
Table 14-9 Drill Summary of Gorumbwa Holes Used in 2017 Mineral Resource Estimate	140
Table 14-10 Gorumbwa Composite Data – 2017 Mineral Resource Estimate	141
Table 14-11 Drill Summary of Pakaka Holes Used in 2017 Mineral Resource Estimate	141
Table 14-12 Pakaka Composite Data – 2017 Mineral Resource Estimate	142
Table 14-13 Drill Summary of Kombokolo Holes Used in December 2017 Mineral Resource Estimate	142
Table 14-14 Kombokolo Composite Data – 2017 Mineral Resource Estimate	143
Table 14-15 Drill Summary of Pamao Holes Used in 2017 Mineral Resource Estimate	144
Table 14-16 Pamao Composite Data – 2017 Mineral Resource Estimate	144
Table 14-17 Mengu Drill Summary of Mengu Holes Used in 2016 Mineral Resource Estimate	145
Table 14-18 Mengu Hill Composite Data – 2016 Mineral Resource Estimate	145
Table 14-19 KCD Open Pit Assigned Density Summary	157
Table 14-20 KCD Underground Assigned Density Summary	158
Table 14-21 Sessenge Assigned Density Summary	159
Table 14-22 Gorumbwa Assigned Density Summary	159
Table 14-23 Pakaka Assigned Density Summary	160
Table 14-24 Kombokolo Assigned Density Summary	160
Table 14-25 Pamao Assigned Density Summary	161
Table 14-26 Mengu Assigned Density Summary	161
Table 14-27 KCD Top Cutting Analysis	167
Table 14-28 Sessenge Top Cutting Analysis	168
Table 14-29 Gorumbwa Top Cutting Analysis	168
Table 14-30 Pakaka Top Cutting Analysis	169
Table 14-31 Kombokolo Top Cutting Analysis	169

18^th September 2018 Page viii

Kibali Gold Mine NI 43-101 Technical Report

Table 14-32 Pamao Top Cutting Analysis	170
Table 14-33 Mengu Top Cutting Analysis	170
Table 14-34 QKNA Parameters for KCD 5003 Domain	182
Table 14-35 QKNA Parameters for Sessenge 9002 Domain	182
Table 14-36 QKNA Parameters for Gorumbwa 1001 Domain	183
Table 14-37 QKNA Parameters for Pakaka 1001 Domain	183
Table 14-38 QKNA Parameters for Kombokolo 1001 Domain	184
Table 14-39 QKNA Parameters for Gorumbwa 2001 Domain	184
Table 14-40 Block Model Variables and Attributes	186
Table 14-41 KCD Global Block Model Extent (With Rotation)	187
Table 14-42 KCD Domain 1001 and 1002 Search Ellipsoid Orientation	187
Table 14-43 Sessenge Global Block Model Extent (No Rotation)	187
Table 14-44 Sessenge Search Ellipsoid Orientation	188
Table 14-45 Gorumbwa Global Block Model Extent (No Rotation)	188
Table 14-46 Gorumbwa Search Ellipsoid Orientation	189
Table 14-47 Pakaka Global Block Model Extent (No Rotation)	189
Table 14-48 Pakaka Search Ellipsoid Orientation	189
Table 14-49 Kombokolo Global Block Model Extent (No Rotation)	190
Table 14-50 Kombokolo Search Ellipsoid Orientation	190
Table 14-51 Pamao Global Block Model Extent (No Rotation)	190
Table 14-52 Pamao Search Ellipsoid Orientation	190
Table 14-53 Mengu Global Block Model Extent (No Rotation)	191
Table 14-54 Mengu Hill Search Ellipsoid Orientation	191
Table 14-55 Kibali Resource Classification Parameters	192
Table 14-56 Block Model Volume Comparison 2017	194
Table 14-57 KCD 2017 Optimisation Parameters	195
Table 14-58 KCD Underground 2017 Optimisation Parameters	195
Table 14-59 KCD Underground 2017 Optimisation Parameters	197
Table 14-60 Sessenge 2017 Optimisation Parameters	202
Table 14-61 Gorumbwa 2017 Optimisation Parameters	202
Table 14-62 Pakaka 2017 Optimisation Parameters	203
Table 14-63 Pakaka Geometallurgical Domained Recoveries	205
Table 14-64 Kombokolo 2017 Optimisation Parameters	206
Table 14-65 Pamao 2017 Optimisation Parameters	206
Table 14-66 Kibali Gold Project Mineral Resource Estimate as of 31st December 2017	209
Table 14-67 KCD Underground 2017 vs 2016 Comparison Within Bounding Box and Within MSO	210
Table 14-68 KCD Underground 2017 versus 2016 Comparison by Classification	211

18^th September 2018 Page ix

Kibali Gold Mine NI 43-101 Technical Report

Table 14-69 Sessenge 2017 vs 2016 Comparison Within $1,500 Pit Shell	211
Table 14-70 Gorumbwa 2017 vs 2016 Comparison Within $1,500 Pit Shell	212
Table 14-71 Pakaka 2017 vs 2016 Comparison Within $1,500 Pit Shell	212
Table 14-72 Kombokolo 2017 vs 2016 Comparison Within $1,500 Pit Shell	213
Table 14-73 Pamao 2017 vs 2016 Comparison Within $1,500 Pit Shell	214
Table 14-74 Mengu 2017 vs 2016 Comparison Within $1,500 Pit Shell	214
Table 14-75 Kibali Mine Call Factor (MCF) 2017 EOY Reconciliation	215
Table 15-1 Total Ore Reserve Estimate at December 31, 2017	222
Table 15-2 Kibali Open Pit Ore Reserves as of 31st December 2017	224
Table 15-3 Open Pit Ore Reserve Comparison to Previous Estimate	225
Table 15-4 Kibali Underground Ore Reserves by Classification and Zone – 31st December 2017	227
Table 15-5 Kibali Underground Ore Reserves by Classification and Zone – 31st December 2017	227
Table 15-6 Underground Ore Reserve Comparison to Previous Estimate	228
Table 15-7 Kibali Surface Stockpile Ore Reserve as of 31 December 2017	229
Table 15-8 KCD Geotechnical Geometry	231
Table 15-9 Rock Mass Properties for the Sessenge Pit	232
Table 15-10 Sessenge Slope Design	233
Table 15-11 Kombokolo Re-Designed Pit Slope Angles	235
Table 15-12 Gorumbwa Recommended Pit Slope Configuration	236
Table 15-13 Pakaka Recommended Pit Slope Configuration	237
Table 15-14 Pamao Recommended Pit Slope Configuration	237
Table 15-15 Summary of 2017 Mining Recovery and Unplanned Dilution	243
Table 15-16 Summary of 2017 Underground Ore Reserve Estimate Dilution Parameters	244
Table 15-17 KCD, Kombokolo, Pakaka Open Pits - Marginal and Full Grade Ore Cut-Off Grade for Different Material Types	246
Table 15-18 Pamao, Sessenge, Gorumbwa Open Pits - Marginal and Full Grade Ore Cut-Off Grade for Different Material Types	247
Table 15-19 Kibali Underground Mine – Cut-Off Grade Calculation	248
Table 15-20 Comparison of Whittle Results for Sessenge Pit with 2016 Results	249
Table 15-21 Comparison of Whittle Results for Pamao Pit with 2016 Results	250
Table 15-22 Sessenge Gold Price Sensitivities	251
Table 15-23 Kombokolo Gold Price Sensitivities	252
Table 15-24 Pamao Gold Price Sensitivities	253
Table 15-25 KCD Gold Price Sensitivities	254
Table 15-26 Sessenge Gold Price versus Pit Size and Sell Price Risk Analysis	256
Table 15-27 Summary of Pit Design Parameters	258

18^th September 2018 Page x

Kibali Gold Mine NI 43-101 Technical Report

Table 16-1 Kibali Open Pits Historical Production	266
Table 16-2 Kibali Open Pits, Reserves Basis	266
Table 16-3 Current Primary Open Pit Mine Equipment Fleet	268
Table 16-4 Waste Dump Capacities	269
Table 16-5 Kibali KCD Underground Historical Production	270
Table 16-6 Kibali Underground Mining Equipment	272
Table 16-7 Mine Plan Material Classification Risk	273
Table 16-8 Open Pits Mining Sequence Over the LOM	275
Table 16-9 Kibali KCD Underground Life of Mine Physicals	277
Table 17-1 2018 Process Feed Plan	285
Table 17-2 Kibali Processing Plant Overall Gold Recovery in 2017 by Month	293
Table 17-3 Kibali Processing Plant Production History	295
Table 17-4 Actual Process and Plant Engineering Operating Costs for 2016 and 2017	298
Table 21-1 LOM Capital Expenditure	325
Table 21-2 LOM Operating Unit Costs	326
Table 21-3 LOM Operating Total Costs	326
Table 25-1 Kibali Risk Analysis	338

18^th September 2018 Page xi

Kibali Gold Mine NI 43-101 Technical Report

List of Figures
Figure 1-1 Simplified Flowsheet of the Kibali Process Plant Depicting Two Discreet Streams	15
Figure 1-2 Initial Hole Composite Dissolutions (Blue Markers = Fresh, Green Markers = Transition, Red Markers = Oxide)	16
Figure 1-3 Graph Indicating Kibali Throughput Performance Against Design of 7.2Mtpa	17
Figure 1-4 Kibali Plant Process Recovery for Year 2017	17
Figure 4-1 Map of the DRC Showing the Location of Haut Uélé District	39
Figure 4-2 Kibali Tenement and Permits	44
Figure 5-1 Kibali 10 Year Rainfall Statistics by Monthly Actual Measurement	46
Figure 5-2 Kibali Deposits and Surrounding Communities	48
Figure 5-3 Kibali Overview Photograph	50
Figure 6-1 Kibali Mineral Resource and Ore Reserve Evolution	55
Figure 7-1 DRC Regional Geology	57
Figure 7-2 Kibali Project Local Interpreted Geology	60
Figure 7-3 Photograph Showing Inter-Bedded Carbonaceous Argillite. Shale and Siltstone with Sub- parallel S0 and S1, Overlain by Transgressive Dissolution Crenulation Cleavage (S3)	61
Figure 7-4 Photograph Showing the Coarse Clastic Conglomerate Unit with Sub-Parallel S0 and S1	61
Figure 7-5 Photograph Showing Pervasive Primary Lithology Texture Destructive Alteration ACSA Which Has Been Cut by Late-Stage Silica and Silica-Pyrite Veins	63
Figure 7-6 Kibali Project Interpreted Geology Illustrating Two Main Mineralisation Trends	65
Figure 7-7 KCD and Sessenge 2017 Block Models with Underground Mine Design	66
Figure 9-1 Kibali Project with Airborne Magnetic Response (TMI – Warm Colours are Magnetic Highs)	74
Figure 9-2 Kibali Project with Airborne EM Response (Channel 7 – Warm colours are Conductive Units)	75
Figure 11-1 Diamond Drill Core Sample Flowchart	85
Figure 11-2 Reverse Circulation Sample Flowchart	86
Figure 11-3 Channel Sample Flowchart	87
Figure 11-4 Kibali QA/QC Protocol Flowchart	90
Figure 11-5 Scatter Plot of CRMs at SGS Doko	93
Figure 11-6 Tram Line Graph for CRMs Analysed at SGS Doko	94
Figure 11-7 Scatter Plot for CRMs Used Between January and December 2017	96
Figure 11-8 Tram Line Graph for CRMs Analysed at SGS Mwanza During the Review Period	97
Figure 11-9 Performance Graph of Blank Samples at SGS Doko During the Review Period	98
Figure 11-10 Performance Graph for Blank Samples at SGS Mwanza	99

18^th September 2018 Page xii

Kibali Gold Mine NI 43-101 Technical Report

Figure 11-11 Ascending Grade Correlation Plot for RC Samples Between 0 g/t Au to 40 g/t Au at SGS Doko	101
Figure 11-12 Precision Plot of RC Field Duplicates vs Original Sample at SGS Doko	102
Figure 11-13 Ascending Grade Correlation Plot for SGS Doko Coarse Duplicates Between 0 g/t Au to 50 g/t Au	103
Figure 11-14 Normal Scatter Plot of SGS Doko Coarse Duplicates £ 50 g/t Au Tail Cut	104
Figure 11-15 Precision Plot of Coarse Duplicates vs Original Sample at SGS Doko	104
Figure 11-16 Ascending Grade Correlation Plot for Mwanza Coarse Duplicates from 0 g/t Au to 60 g/t Au at SGS Mwanza	105
Figure 11-17 Normal Scatter Plot of Duplicates for Data at 60 g/t Au Tail Cut at SGS Mwanza	106
Figure 11-18 Precision Plot of Pulp Duplicates vs Original Sample at SGS Mwanza	106
Figure 11-19 Ascending Grade Correlation Plot for SGS Doko Trench Field Duplicates Between 0 g/t Au to 10 g/t Au	108
Figure 11-20 Normal Scatter Plot of SGS Doko Trench Field Duplicates £ 10 g/t Au Tail Cut	108
Figure 11-21 Precision Plot of Trench Field Duplicates vs Original Sample at SGS Doko	109
Figure 11-22 Normal QQ Plot of Pulp Duplicates SGS Doko vs. OMAC at 10 g/t Au Tails Cut	110
Figure 13-1 Initial Hole Composite Dissolutions (Blue Markers = Fresh, Green Markers = Transition, Red Markers = Oxide)	119
Figure 13-2 Primary Extraction Excluding the Leaching of Flotation Tails	121
Figure 13-3 Primary Extraction Variability Including the Leaching of the Flotation Tails	123
Figure 13-4 Extraction as a Function of Diamond Drill Holes	124
Figure 13-5 Analysis of the Drill Hole Samples Exhibiting Large Variances	124
Figure 13-6 Gold Extractions Obtained for Various Extraction Variability Tests and Master Composite Samples (OMC report)	125
Figure 13-7 Plots of Extraction Using the Primary Process for the Oxide Materials – KCD	126
Figure 13-8 Direct Cyanidation of Grade Control Samples	126
Figure 13-9 2016 Sessenge Geomet Model	128
Figure 13-10 Sampling Strategy and Classification of Samples in the KCD	130
Figure 14-1 Boundary Analysis between HG (5101) and LG (5005) Domains from KCD	147
Figure 14-2 3D View of KCD Sessenge Mineralisation Lodes (3000 Lodes in Orange, 5000 in Red, 9000 in Pink) and $1,000 Pit Design in Grey	148
Figure 14-3 3D View of Gorumbwa Mineralisation (Yellow) Depletion Model (Magenta) and $1,000 & $1,500 whittle Pit Shells	151
Figure 14-4 3D View Pakaka Low-grade Mineralisation (Grey), High-grade Mineralisation (Orange) and Optimised Pit Shells ($1,000 Dark Blue & $1,500 Light Blue)	152
Figure 14-5 3D View of Kombokolo Mineralisation Within Both $1,000 & $1,500 Whittle Shells	153
Figure 14-6 3D View of Pamao Mineralisation	154
Figure 14-7 3D View of Mengu Mineralisation and Pit Shells	155

18^th September 2018 Page xiii

Kibali Gold Mine NI 43-101 Technical Report

Figure 14-8 KCD Log Histogram and Log Probability Plot of 2 m Uncapped Composites Within Mineralised Lodes	163
Figure 14-9 Sessenge Log Histogram and Log Probability Plot of 2 m Uncapped Composites Within Mineralised Lodes	163
Figure 14-10 Gorumbwa Log Histogram and Log Probability Plot of 2 m Uncapped Composites Within Mineralised Lodes	164
Figure 14-11 Pakaka Log Histogram and Log Probability Plot of 2 m Uncapped Composites Within Mineralised Lodes	164
Figure 14-12 Kombokolo Log Histogram and Log Probability Plot of 2 m Uncapped Composites Within Mineralised Lodes	165
Figure 14-13 Pamao Log Histogram and Log Probability Plot of 2 m Uncapped Composites Within Mineralised Lode	165
Figure 14-14 Mengu Log Histogram and Log Probability Plot of 2 m Uncapped Composites Within Mineralised Lodes	166
Figure 14-15 KCD 3102 Normal Score Variogram Models	172
Figure 14-16 KCD 3102 Nested Back Transformed Variogram Model	172
Figure 14-17 KCD 5101 Normal Score Variogram Models	173
Figure 14-18 KCD 5101 Nested Back Transformed Variogram Model	173
Figure 14-19 KCD 9105 Nested Back Transformed Variogram Model	174
Figure 14-20 KCD 9105 Nested Back Transformed Variogram Model	174
Figure 14-21 Sessenge 9002 Normal Score Variogram Models	175
Figure 14-22 Sessenge 9002 Nested Back Transformed Variogram Model	175
Figure 14-23 Gorumbwa 1001 Normal Score Variogram Models	176
Figure 14-24 Gorumbwa 1001 Nested Back Transformed Variogram Model	176
Figure 14-25 Pakaka 1001 Normal Score Variogram Models	177
Figure 14-26 Pakaka 1001 Nested Back Transformed Variogram Model	177
Figure 14-27 Kombokolo 1101 and 1002 Normal Score Variogram Models	178
Figure 14-28 Kombokolo 1101 and 1002 Nested Back Transformed Variogram Model	178
Figure 14-29 Pamao Normal Score Variogram Models	179
Figure 14-30 Pamao Nested Back Transformed Variogram Model	179
Figure 14-31 Mengu Hill 1001 Normal Score Variogram Models	180
Figure 14-32 Mengu Hill 1001 Nested Back Transformed Variogram Model	180
Figure 14-33 QKNA for Sessenge Domain 9103 Open Pit GC Zone	181
Figure 14-34 KCD SWATH Plot of Domains 5102 Along Y Axis	194
Figure 14-35 KCD 3D View of MSO Shapes Against Underground Blocks Above 1.6 g/t Au – View Towards SE	198
Figure 14-36 KCD 3D View of MSO Shapes Against Grade Blocks – View Towards SE	199
Figure 14-37 KCD 3D View of MSO Shapes Against Grade Blocks – View Towards SE	200

18^th September 2018 Page xiv

Kibali Gold Mine NI 43-101 Technical Report

Figure 14-38 KCD 3D Vew of MSO Shapes Against Grade Blocks – View Towards SE	201
Figure 14-39 Plan View Map of the Pakaka Geometallurgical Domains and Their Spatial Correlation with the Mineralisation Resource Domains	204
Figure 14-40 2017 Kibali Mine Production with Weekly Feed Source Ratio versus Pulp Call versus Gold after Smelting	216
Figure 14-41 2017 Weekly Grades Comparison (GC Call Grade vs Plant Check Out Grade vs Carbon Loading)	216
Figure 14-42 2017 Weekly Tonnage Comparison (GC Call Tonnes vs Plant Check Out Tonnes)	217
Figure 15-1 KCD Underground Mining Zones	223
Figure 15-2 Kibali Underground Ore Reserve Classification (View NW)	226
Figure 15-3 History of Kibali Underground Measured + Indicated Mineral Resources and Ore Reserves	228
Figure 15-4 KCD Pushback 3 Voids Model Within the Planned Pit	229
Figure 15-5 Gorumbwa Sections of Completed Voids	230
Figure 15-6 KCD Pushback 3 Geotechnical Domains	231
Figure 15-7 Sessenge Geotechnical Domains	232
Figure 15-8 Long Section of Kombokolo Pit Showing Deeper Weathering Profile	233
Figure 15-9 Geotechnical Domains for Kombokolo Pit	234
Figure 15-10 Geotechnical Domains for Gorumbwa Pit	235
Figure 15-11 Geotechnical Domains for Pakaka Pit	236
Figure 15-12 Layout of Dewatering Boreholes in Sessenge Pit	238
Figure 15-13 Section of Gorumbwa Drawdown in Old Voids	239
Figure 15-14 KCD Sensitivity Analysis	255
Figure 15-15 Sessenge Value/Tonnage Curve	257
Figure 15-16 Pakaka Pushback Designs	259
Figure 15-17 KCD End of 2017 Reserve Pit	259
Figure 15-18 Sessenge E End of 2017 Reserve Pit	260
Figure 15-19 Pakaka End of 2017 Reserve Pit	260
Figure 15-20 Kombokolo End of 2017 Reserve Pit	261
Figure 15-21 Pamao End of 2017 Reserve Pit	261
Figure 15-22 Gorumbwa End of 2017 Reserve Pit	262
Figure 15-23 Life of Mine Development and As Built (Dec 2017)	263
Figure 15-24 Materials Handling System	264
Figure 16-1 Plan Showing Relative Positions of Open Pits and Main Mine Infrastructure	265
Figure 16-2 Long Section of the KCD Pit and Pushback 3	267
Figure 16-3 Long Section of the Pakaka Pit and Pushback 2	267
Figure 16-4 Long Section of the Gorumbwa Pit and Pushback 2	268

18^th September 2018 Page xv

Kibali Gold Mine NI 43-101 Technical Report

Figure 16-5 KCD Underground Mining Zones	270
Figure 16-6 Long Hole Open Stoping Methods	271
Figure 16-7 Historic Rainfall Pattern and Lost Production Hours from Rain	273
Figure 16-8 Kibali Open Pit Mining Rate	274
Figure 16-9 KCD Underground LOM on 1st Jan 2019	278
Figure 16-10 KCD Underground LOM on 1st Jan 2021	278
Figure 16-11 KCD Underground LOM on 1st Jan 2023	279
Figure 16-12 KCD Underground LOM on 1st Jan 2025	279
Figure 16-13 KCD Underground LOM on 1st Jan 2027	280
Figure 16-14 KCD Underground LOM on 1st Jan 2035	280
Figure 17-1 Simplified Flowsheet of the Kibali Process Plant Depicting Two Discreet Streams	284
Figure 17-2 Kibali Plant Performance – Tonnes Treated 2013 to 2017	285
Figure 17-3 Kibali Processing Plant Overall Gold Recovery in 2017	292
Figure 17-4 Kibali Plant Recovery	294
Figure 17-5 Kibali Plant Pumpcell Residue and Throughput	294
Figure 17-6 Kibali Plant Production History	296
Figure 17-7 Kibali Concentrate Storage Facility	297
Figure 17-8 Kibali Processing Weekly Twin Stream Sulphide Proportion	298
Figure 18-1 Kibali Water Management Plan	303
Figure 18-2 Kibali Electrical Supply Mix	305

18^th September 2018 Page xvi

Kibali Gold Mine NI 43-101 Technical Report

1 Executive Summary

The purpose of this report is to support the public disclosure of the year end 2017 Mineral Resource and Ore Reserve estimates at the Kibali Gold Mine (Kibali, the Mine or the Project) located in the Democratic Republic of the Congo (DRC) as of 31st December 2017. This Technical Report conforms to National Instrument 43-101 - Standards of Disclosure for Mineral Projects (NI 43-101). All currency in this report is US dollars ($) unless otherwise noted.

Kibali Goldmines SA (Kibali Goldmines) is an exploration and mining company which is currently owned 70% by Moto Gold Mines Limited (Moto), and 20% by Kibali Jersey Ltd. Both Moto and Kibali Jersey Ltd. are joint ventures owned 50% by Randgold Resources Limited (Randgold) and 50% AngloGold Ashanti Limited (AngloGold Ashanti). This equates to an overall interest in Kibali Goldmines of 45% for Randgold and 45% for AngloGold Ashanti. The remaining 10% of Kibali Goldmines SA is owned by Congolese parastatal Société Miniere de Kilo-Moto SA UNISARL (SOKIMO) with the shareholding held by the Minister of Portfolio of DRC (MoP).

Randgold is the operator at Kibali for both exploration and mining. In addition, Randgold is developing and operating gold mines in West and East Africa. The most notable of these are the following:

o Morila Gold Mine in Mali,

o Loulo Gold Mine in Mali,

o Gounkoto Gold Mine in Mali,

o Tongon Gold Mine in the Ivory Coast, and

o Massawa Exploration Project in Senegal.

The Project is an operating mine consisting of the Kibali Karagba-Chauffeur-Durba (KCD) underground mine, the KCD open pit, satellite deposits, a processing plant (7.2 Mtpa capacity), three hydropower stations, together with other associated mine operation and regional exploration infrastructure. The plant produces gold doré bars.

Total mine production from underground and open pits in 2017 was 7.6 Mt at a head grade of 2.9 g/t Au for a total of 596 koz gold (83.4% recovery). The shaft infrastructure was completed and commissioned in 2017. Underground mining continued to ramp-up during the year, which accounted for 1.8 Mt of this production.

Total production since mining commenced to year end 2017 is 161 Mt (30 Mt ore) for 2.4 Moz of gold (82.3% recovery).

The effective date of this report is 31st December 2017.

18^th September 2018 Page 1

Kibali Gold Mine NI 43-101 Technical Report

1.1 Property Description and Location

The Kibali Mine is a gold mining and exploration project, which is located in the NE of the Democratic Republic of Congo (DRC), approximately 560 km NE of the city of Kisangani and 150 km west of the Ugandan border town of Arua, near the international borders with Uganda and Sudan. Kinshasa, the capital city of DRC, is located approximately 1,800 km SW of the Project.

Personnel access to the Project is commonly through charter flight directly to site from Entebbe, Uganda which is served daily by commercial flights from European cities.

Road access is available from Kampala, Uganda and is approximately 650 km, which provides the primary route for operational supply chain.

The Project covers an area of approximately 1,836 km², is centred at approximately 3.13° north and 29.58° east, is situated within two territories, namely Watsa and Faradje which fall under the administrative district of Haut Uélé.

The Project consists of multiple mineral deposits including; Karagba-Chauffeur-Durba (KCD), Sessenge, Sessenge SW, Pakaka, Pamao, Gorumbwa, Mengu Hill, Mengu Village, Megi, Marakeke, and Kombokolo.

1.2 Mineral Rights and Land Ownership

Kibali Goldmines SA has been granted ten Exploitation (Mining) Permits under the DRC Mining Code (2002) in respect of the Kibali Gold Project, eight of which are valid until 2029 and two of which are valid until 2030. All Mineral Resources and Ore Reserves summarised in this report are contained within these Permits.

The principle mineral deposit, KCD, is formed as both an open pit and underground mine. This operation and the associated infrastructure (processing plant, accommodation, and airport) are within Exploitation Permits 11447 and 11467. All Kibali Exploitation Permits are presented in Table 1-1.

18^th September 2018 Page 2

Kibali Gold Mine NI 43-101 Technical Report

Table 1-1 Kibali Exploitation Permit Details

Arête No.	Permit No.	Surface Area       (km2)	Expiry Year
0852/CAB.MIN/MINES/01/2009	11447	226.8	2029
0855/CAB.MIN/MINES/01/2009	11467	248.9	2029
0854/CAB.MIN/MINES/01/2009	11468	45.9	2030
0853/CAB.MIN/MINES/01/2009	11469	91.8	2029
0329/CAB.MIN/MINES/01/2009	11470	30.6	2029
0852/CAB.MIN/MINES/01/2009	11471	113.0	2029
0331/CAB.MIN/MINES/01/2009	11472	85.0	2029
0856/CAB.MIN/MINES/01/2009	5052	302.4	2029
0858/CAB.MIN/MINES/01/2009	5073	399.3	2029
0103/CAB.MIN/MINES/01/2011	5088	292.2	2030

The Kibali operation conforms to the DRC Mining Code (2002) and regulations. In the Qualified person’s opinion, all appropriate Permits have been acquired and obtained to conduct the work proposed for the property.

The next renewal dates for the Permits are 5^th November 2029 and 6^th March 2030 and the current life of mine plan (LOM) for the Kibali Ore Reserves extends beyond these dates.

The DRC Mining Code (2002) includes provision for renewal of all Exploitation Permits for a successive period of 15 years, providing the holder has not breached the Permit obligations of Permit fee and annual surface rights fees payment and upholds all environmental standards set out in the Exploitation Permit. Furthermore, the Permit holder should provide the appropriate government departments with a monthly mining activity report and quarterly exploration reports.

All the Permit fees, surface rights fees, and taxes relating to Kibali Goldmines exploitation rights have been paid to date and the concession is in good standing.

The DRC Mining Code (2002) and Regulations have been amended with an updated Mining Code which came into force on 9th March 2018 (DRC Mining Code(2018) and the related amended Mining Regulations which came into force on 8th June 2018.

The following changes have been made to the DRC Mining Code (2002) that could have an impact on Kibali:

☐ Royalty charges are to be increased from 2.5% to 3.5%. This increases royalty charges over the LOM by an estimated $94.5 M, which would not materially impact the LOM profitability.

☐ Various increases in import and other duties from 4% to 7% depending on consumable type, which would not materially impact the LOM profitability.

☐ A super-tax profit has been promulgated based on the feasibility study prepared at the time the approval was given for the building of the Project. Accordingly, such a tax would only apply if the average annual gold price was in excess of $2,000/oz.

The exact impact, if any, of the changes will only be fully known once the 2018 Mining Code and related regulations are clarified and implemented in full.

Going forward the DRC Mining Code (2018) envisages a stability period for the tax, customs, and exchange control regime of five years from the date on which the DRC Mining Code (2018) came

18^th September 2018 Page 3

Kibali Gold Mine NI 43-101 Technical Report

into force and further provides that a number of the taxes shall be applied in accordance with the applicable substantive law.

Kibali Jersey Limited, the holding company of Kibali, the shareholders of Kibali Jersey Limited and Kibali Gold Mines SA, are considering all options to protect their vested rights under the DRC Mining Code and to enforce the additional state guarantees previously received, including preparations for international arbitration. In addition, engagement with the DRC government is ongoing, with the aim of exploring alternative solutions, which could be mutually acceptable to both parties. This includes the application of Article 220 of the DRC Mining Code (2018), which affords benefits to mining companies such as Kibali, operating in landlocked infrastructurally challenged provinces. If Article 220 were applied to Kibali, any advantages granted would mitigate any impact of the implementation of the DRC Mining Code (2018).

The Qualified Person notes that the mining code is in the process of transition and the current proposed changes do not have an impact on the stated Mineral Resource or Ore Reserves at the gold prices used for Ore Reserve ($1,000/oz), Mineral Resource ($1,500/oz), or the current gold prices of $1,200/oz (August 2018).

In the Qualified Person’s opinion, all appropriate permits have been acquired and obtained to conduct the work proposed for the property.

The Qualified Person is not aware of any risks that could result in the loss of ownership of the deposits or loss of the permits, in part or in whole.

The Qualified Person is not aware of any other significant factors and risks that may affect access, title, or the right of ability to perform work on the property.

1.3 History

The discovery of gold in the NE of the Democratic Republic of Congo (DRC) is attributed to Hannan and O’Brien in 1903.

Historical gold production from the Kilo and Moto areas between 1906 and 2009 is estimated to be approximately 11 Moz, half of which came from alluvial deposits. Mining operations were conducted by the Belgian Government via the Société des Mines d’Or de Kilo-Moto (SOKIMO), which was established in 1926. Most of the mining activity within the Project area was undertaken during the 1950s but accurate production records have been lost over the years of civil unrest in the region. Gorumbwa, Agbarabo and Durba deposits are believed to have produced more than 60% of the over 3 Moz of recorded gold production from the Moto area.

After independence in 1960, gold production dropped sharply as mining was mainly undertaken by artisanal workers and small-scale alluvial operations. SOKIMO changed its name to Offices des Mines d’Or de Kilo-Moto (OKIMO) in 1966 and was the main operator in the Project area. Sporadic underground mining was conducted in the Project area after 1960, however this is believed to be of a remnant nature and as such negligible amounts of gold were produced. Accurate production records are not available due to the civil unrest in the region during the 1980s and 1990s.

18^th September 2018 Page 4

Kibali Gold Mine NI 43-101 Technical Report

The KCD deposit was originally discovered by a joint venture (JV) between Barrick and AngloGold Ashanti in 1998. The Barrick and AngloGold Ashanti JV completed a number of drilling programs, mainly concentrated at KCD and Pakaka. AngloGold Ashanti and Barrick withdrew from the Project in 1998 due to local unrest and civil war.

Moto Goldmines Limited (Moto) acquired the available 70% stake in the Project in 2004. Moto completed a pre-feasibility study in 2006, a Feasibility Study in December 2007, and an Optimised Feasibility Study in March 2009.

In July 2009, Randgold and AngloGold Ashanti entered into a 50/50 JV, which acquired Moto and their 70% ownership of the Project. In December 2009, the JV acquired an additional 20% shareholding of Kibali Goldmines from SOKIMO. The DRC State remained a partner in the Project through OKIMO retaining a 10% interest.

Kibali Goldmines undertook a feasibility study update which doubled the declared Ore Reserve to over 10 Moz of gold. Subsequently, construction was approved in 2012 and the Kibali mine has been developed in two key phases.

Phase 1 encompassed the construction and mining of KCD open pit operation which commenced mining in July 2012, together with construction of the processing plant and commissioning of the oxide processing circuit began in the third quarter of 2013. Kibali poured its first gold in September 2013, ahead of plan, and started commercial production in the fourth quarter of 2013. A 36-unit high speed thermal power station was constructed to support the power generation requirements of the mine, together with the first of three hydropower stations Nzoro 2. This first phase of development was completed in December 2014.

Phase 2 comprises construction of the underground mine development, including the vertical shaft and twin declines, in addition to the associated Project infrastructure to support mining of other satellite open pit operations including Mengu-Hill, Pakaka, Kombokolo and Rhino. During this phase two additional hydropower stations were constructed and commissioned, the first of which Ambarau was commissioned at the start of 2017 and the second Azambi is scheduled for commissioning in 2018, along with further satellite pit developments including Sessenge and Gorumbwa.

Commissioning of the sulphide circuit began early in 2014 and production has steadily ramped-up since then with the mine now consistently exceeding its processing design capacity.

1.4 Geology and Mineralisation

The Kibali deposits are hosted within the Kibali Greenstone Belt (otherwise referred to as Moto granite-greenstone terrane), bounded to the north by the West Nile Gneiss and to the south by plutonic rocks of the Watsa district. The Kibali Greenstone Belt is an elongate WNW-ESE trending terrane containing Archean aged volcano-sedimentary conglomerate, carbonaceous shales, siltstone, banded iron formations, sub aerial basalts, mafic intermediate intrusions (dykes and sills) and multiple intrusive phases that range from granodiorite, to gabbroic in composition. Based on

18^th September 2018 Page 5

Kibali Gold Mine NI 43-101 Technical Report

textures and types of lithologies present in the stratigraphy, the rocks within the Project area are interpreted as having been laid down in an aqueous environment.

The majority of the primary lithologies are clastic (sedimentary) in origin, possibly being developed in a regional extensional environment such as a rift graben or half graben. At Kibali, the gold deposits are largely hosted in siliciclastic rocks, banded iron formations, and cherts that were metamorphosed under greenschist facies conditions, situated along a curvilinear zone 20 km long and up to one km in width, known as the “KZ Structure”. Gold mineralisation is concentrated in gently NE to NNE-plunging fold axes whose orientations are generally parallel with a prominent lineation in the mineralised rocks.

The Kibali deposits differ from many orogenic gold deposits as they are hosted within a thrust stack sequence with ductile to brittle-ductile deformational structures and a complex folding history. There are two principal structure sets: NW-SE striking, NE dipping thrust faults and a series of sub-vertical NE-SW shear structures both of which in association with the folding are considered important mineralising controls. Unlike many other orogenic gold deposits, mineralisation within the Kibali district typically lacks significant phases of quartz-rich veins.

The mineralised deposits of the Kibali district are associated with halos of quartz, ankerite, and sericite (ACSA-A) alteration that extend for 10s to 100s of metres into the adjacent rocks. Areas of economic mineralisation are defined where the project scale ACSA-A alteration is locally overprinted by the ankerite-siderite, pyrite alteration assemblage (ACSA-B) that hosts the gold mineralisation. The gold bearing sulphides consist of disseminated pyrite, minor pyrrhotite, and arsenopyrite. The auriferous pyrite occurs as both ‘salt and pepper’ disseminated fine grains and bleb-like clusters of disseminated grains.

The KCD deposit is the principal mineralised occurrence along the Sessenge-KCD Trend. It consists of three semi-vertically stacked lodes hosted within the volcano-sedimentary units, conglomerate units, and ironstone and chert assemblages. The location of the individual lodes within the KCD deposit are intimately controlled by the position, shape, and orientation of a series of gently NE-plunging fold axes. The lodes may be linked genetically by large-scale recumbent folding developed between two bounding NE trending structures.

Higher grade zones of strong to intense alteration overprint and texturally-destructive foliation and lithological textures. These are broadly categorised as the 3000 lodes, 5000 lodes, and the 9000 lodes, all of which plunge towards the NE at low to moderate angles (approximately 30°) with drilling intercepts indicating a down plunge continuation of approximately 2,000 m (remaining open down plunge).

The 3000 lode crops out in the present open pit (Karagba) and is the western-most lode. It is approximately 300 m in width, 30 m thick, and has a broad gentle and open semi-synclinal form to its plunge. The 5000 lode outcrops slightly east and south of the 3000 lode (Chauffeur and Durba) and forms the majority of the topographically elevated area known as Durba Hill. The lodes are more sub-vertical in attitude than the 3000 and 9000 lodes and are of a consistently higher grade. The 9000 lode outcrops out to the south of Durba Hill, forming the Mineral Resources in the Sessenge open pit. The 9000 lode is comprised of two main lodes; 9101 and 9105. The 9105 lode is of a similar

18^th September 2018 Page 6

Kibali Gold Mine NI 43-101 Technical Report

shape and attitude as the 5000 lode, with the flat underlying 9101 lode extending up-plunge to Sessenge open pit mineralisation.

Both Gorumbwa and Kombokolo deposits occur along a NE trending mineralised corridor located 800 m to the west of the main Sessenge-KCD structural zone. Both are considered to be formed from the same mineralising event, with similar alteration and structural characteristics to the KCD deposit but significantly smaller in size. The Gorumbwa deposit was mined by SOKIMO in 1955 from both underground and via a small open pit operation, with total production estimated to have been approximately 2.8 Mt at 7 g/t Au. The underground and open pit workings are presently collapsed and flooded. The mineralisation consists of a series of stacked ‘lenses’ that variably extend down plunge for a length of 1,000 m at an average width of 200 m and which have been mined to a depth of 400 m below topographic surface.

The Mengu Hill deposit lies near the NW end of the NW trending Pakaka-Mengu Trend. The stratigraphy in the vicinity of the deposit is dominated by a meta-conglomerate unit that is interbedded with fine-grained sediments, siliceous sericite schist and minor mafic volcanic rocks. These lithologies overlay a massive magnetite and specular hematite ironstone-chert unit that has weathered to create the topographic high, namely “Mengu Hill”. Mineralisation is associated with silica-ankerite-pyrite alteration and is focused within the ironstone unit and along its contact with the overlying conglomerate unit. The mineralised lens is cigar like in shape and plunges shallowly to the NE with development of refolded folds or sheaths. The Mengu Hill mineralisation averages 150 m in width and continues 700 m down plunge to a depth of 250 m below the topographical surface.

The Pakaka-Pamao deposits are located at the SE end of the 7 km NW trending Pakaka-Mengu Trend. Gold mineralisation at Pakaka-Pamao is hosted by the meta-conglomerate interbedded with minor tuffaceous units. Recent works show mineralisation to be hosted in meta-sandstone and banded iron formation. The mineralised zones are characterised by silica-ankerite-pyrite alteration, mainly in well-foliated siliceous rocks. Higher gold grades appear to correlate well with the presence of abundant pyrite and arsenopyrite, and spatially associated with the intersection of the NW trending thrust surface and a NE trending strain corridor. The structures combine to produce a broad NE plunging open anticlinal structure, with Pamao on the west limb and Pakaka on the east. The Pakaka mineralisation continues down plunge beyond the limits of the drilling and represents further exploration potential. The Pakaka mineralisation extends over a strike length of 1,000 m, averages a thickness of 30 m and has been identified to at least a depth of 350 m below surface.

Mengu Village is located near the NW end of the Pakaka-Mengu Trend. The mineralisation is tabular in form, trending NW and dipping shallowly to the NE and is hosted by conglomerates with thin ironstone and carbonaceous shale intercalations. The mineralisation is approximately 150 m in strike length with an average thickness of 15 m and has been identified to a depth of 150 m below the surface.

The Marakeke deposit is located midway along the Pakaka-Mengu Trend with mineralisation developed in a variably carbonate-sericite-silica altered ironstone-chert, that dips to the NE at approximately 30° and strikes WNW. The Marakeke deposit occurs as a single tabular lens typically between 10 m to 30 m thick that trends NW and dips gently to the NE. The mineralised zone has

18^th September 2018 Page 7

Kibali Gold Mine NI 43-101 Technical Report

been tested by drilling over a strike length of approximately 1,000 m and extends some 200 m down dip.

1.5 Exploration

The Project has been explored since the early 1900s by SOKIMO and more recently by Barrick and AngloGold Ashanti and then Moto using geochemistry sampling, mapping, trenching, geophysical surveys, and drilling. Kibali Goldmines has been exploring at Kibali since 2010.

Exploration at Kibali focusses on advancing both brownfields and greenfields targets. Brownfields exploration involves testing underground and open pit targets for extensions of high-grade mineralisation based on the structural model, but commonly in a down plunge direction as the major axis of continuity.

Satellite deposits and gaps between existing Mineral Resources are evaluated by exploration work to define Mineral Resources from conceptual targets. During 2018, a key exploration programme will target the previously identified prospect of Kalimva-Ikamva with the aim of defining Inferred Mineral Resources.

The 2018 resource definition exploration is scheduled to target the down plunge extension of the KCD 5000 lodes focussing above the bottom level of the shaft, with drilling from a dedicated underground exploration drill drive, whilst continuing the advanced grade control programme ahead of development and infill grade control programme for final pre-production definition of Measured Resources.

1.6 Mineral Resources

The Mineral Resource estimates have been prepared according to the of the guidelines Australasian Code for Reporting of Exploration Results, Mineral Resources and Ore Reserves standards and guidelines published and maintained by the Joint Ore Reserves Committee of the Australasian Institute of Mining and Metallurgy and the Australian Institute of Geoscientists and Minerals Council of Australia (the JORC (2012) Code). Randgold has reconciled the Mineral Resources and Ore Reserves to Canadian Institute of Mining, Metallurgy and Petroleum (CIM) 2014 Definition Standards for Mineral Resources and Mineral Reserves dated May 10, 2014 (CIM (2014) Standards) as incorporated with NI 43-101 and there are no material differences.

Quality assurance and quality control (QA/QC) has been undertaken across the life of exploration to minimise errors. A standard operating procedure (SOP) outlines Kibali Goldmines approach to QA/QC which meets industry best practice. The results from the 2017 QA/QC program show that the performance of the SGS Doko laboratory is meeting industry standards.

For the year end 2017 Mineral Resource update, KCD, Sessenge, Gorumbwa, Kombokolo, Pamao, and Pakaka models within the open pit portions have been updated. These updates were required as a result of new drilling data that was added through both grade control and resource definition drilling along with void model completed in Gorumbwa.

18^th September 2018 Page 8

Kibali Gold Mine NI 43-101 Technical Report

Total Mineral Resources for Kibali are estimated to be 126 Mt at an average grade of 3.26 g/t Au for 13 Moz Au in the Measured and Indicated categories and 44 Mt at an average grade of 2.3 g/t Au for 3.3 Moz Au in the Inferred category. A summary of the Kibali Mineral Resources is presented in Table 1-2. These Mineral Resources have been depleted to 31^st December 2017 using the mined-out surfaces and voids.

Table 1-2 Mineral Resource Statement for the Kibali Mine as of 31^st December 2017

Type	Category	Tonnes (Mt)	Grade     (Au g/t)	Contained     Gold (Moz)	*Attributable     Gold (Moz)
Stockpiles	Measured	1.7	1.45	0.080	0.036
Open Pits	Measured	8.6	2.63	0.73	0.33
	Indicated	39	2.11	2.6	1.2
	Inferred	22	1.8	1.3	0.59
Underground	Measured	12	5.57	2.1	0.96
	Indicated	65	3.64	7.6	3.4
	Inferred	22	2.8	2.0	0.91
Total Mineral Resources	Measured	22	4.11	3.0	1.3
	Indicated	104	3.07	10	4.6
	Measured and Indicated	126	3.26	13	5.9
	Inferred	44	2.3	3.3	1.5

*Attributable Gold (Moz) refers to the quantity attributable to Randgold based on Randgold’s 45% interest in the Kibali Goldmines.

The Mineral Resource estimate has been prepared according to JORC (2012) Code. Kibali have reconciled the Mineral Resources to CIM (2014) Standards, and there are no material differences.

All Mineral Resources tabulations are reported inclusive of that material which is then modified to form Ore Reserves.

Open pit Mineral Resources are Mineral Resources within the $1,500/oz pit shell reported at an average cut-off grade of 0.6 g/t Au. Underground Mineral Resources in the KCD deposit are Mineral Resources, which meet a cut-off grade of 1.6 g/t Au and are reported insitu within a minimum mineable stope shape, at a gold price of $1,500/oz.

Mineral Resources were estimated by Simon Bottoms, CGeol, an officer of the company and Qualified Person. Numbers may not add due to rounding.

The cut-off grade selected for limiting each of the Mineral Resources corresponds to the insitu marginal cut-off grade using a gold price of $1,500/oz.

For the open pit Mineral Resources, the pit shell selected for limiting each of the Mineral Resources corresponds to a gold price of $1,500/oz. As a result of the optimisation process, this pit shell selection will result in the highest undiscounted net present value of the deposit, at $1,500/oz.

Underground Mineral Resources were reported within a minimum mineable stope shape, applying reasonable mineability constraints, including a minimum mining width, a reasonable distance from current or planned development, and a measure of assumed profitability at the related resource cut-off grade, thus deemed as having a reasonable prospect of eventual economic extraction.

The Qualified Person is not aware of any environmental, permitting, legal, title, socioeconomic, marketing, metallurgical, taxation or other relevant factors, which could materially affect the Mineral Resource estimate.

18^th September 2018 Page 9

Kibali Gold Mine NI 43-101 Technical Report

1.7 Ore Reserves

The Ore Reserve estimates have been prepared according to the guidelines Australasian Code for Reporting of Exploration Results, Mineral Resources and Ore Reserves standards and guidelines published and maintained by the Joint Ore Reserves Committee of the Australasian Institute of Mining and Metallurgy and the Australian Institute of Geoscientists and Minerals Council of Australia (the JORC (2012) Code). Randgold has reconciled the Mineral Resources and Ore Reserves to Canadian Institute of Mining, Metallurgy and Petroleum (CIM) 2014 Definition Standards for Mineral Resources and Mineral Reserves dated May 10, 2014 (CIM (2014) Standards) as incorporated with NI 43-101 and there are no material differences.

The Ore Reserve estimates use updated economic factors, the latest resource and geological models, geotechnical inputs, and the latest metallurgical updates. Some inputs were shared across all the operations during the preparation of the Ore Reserve estimates. Ore Reserves were based on the development of appropriately detailed and engineered life of mine (LOM) plans. All design and scheduling work was undertaken to a suitable level of detail by experienced engineers using mine planning software. The planning process incorporated appropriate modifying factors and the use of cut-off grades and other technical-economic investigations. Ore Reserves are stated:

☐ As of 31^st December 2017

☐ At a gold price of $1,000/oz

☐ As Run-of-Mine (ROM) grades and tonnage as delivered to the plant

☐ Including only Measured and Indicated Mineral Resources.

The year end 2017 (100% basis), the total Proved and Probable Ore Reserves are estimated to be 66 Mt at 4.09 g/t Au, containing 8.7 Moz of gold of which 3.9 Moz are attributable to Randgold.

The Kibali Mine Ore Reserve Statement as of 31st December 2017 is shown in Table 1-3.

18^th September 2018 Page 10

Kibali Gold Mine NI 43-101 Technical Report

Table 1-3 Kibali Mine Ore Reserve Summary as of 31^st December 2017

Type	Category	Tonnes (Mt)	Grade     (Au g/t)	Contained Gold (Moz)	*Attributable Gold (Moz)
Stockpiles	Proved	1.7	1.45	0.080	0.036
Open Pits	Proved	4.9	2.72	0.43	0.19
	Probable	16	2.28	1.2	0.54
Underground	Proved	12	4.97	2.0	0.89
	Probable	31	5.06	5.0	2.3
Total Ore Reserves	Proved	19	4.07	2.5	1.1
	Probable	47	4.10	6.2	2.8
	Proved and Probable	66	4.09	8.7	3.9

*Attributable Gold (Moz) refers to the quantity attributable to Randgold based on Randgold’s 45% interest in the Kibali Gold Mines. Ore Reserves are reported on a 100% and attributable basis.

The Ore Reserve estimate has been prepared according to JORC (2012) Code. The Qualified Person has reconciled the Ore Reserves to CIM (2014) Standards, and there are no material differences.

Open pit Ore Reserves are reported at a gold price of $1,000/oz, except for the KCD pit at $1,100/oz, and an average cut-off grade of 1.0 g/t Au including dilution and ore loss factors.

Underground Ore Reserves are reported at a gold price of $1,000/oz and a cut-off grade of 2.5 g/t Au including dilution and ore loss factors.

Open pit and underground Ore Reserves were estimated by Mr. Andrew Law, HND(MMin), MBA, FAusIMM (CP), FIQA, MAICD, AFAIM, an external consultant and Qualified Person.

Numbers may not add due to rounding.

The year end 2017 Ore Reserve estimate shows a net reduction of 0.49 Moz when compared to the estimate for year end 2016. This is mainly due to mining depletion, compensated by some positive model changes resulting from infill grade control drilling and various adjustments to economic parameters.

The average dilution for the underground deposits ranged from 1% to 9%, with mining losses estimated at 3% except for some secondary stopes in 5101 and 5102 which were estimated at 10%. The average dilution for the open pit deposits was 11.3% and ranges between 9.6% and 13%. Mining losses were estimated at 3% except for larger void areas in the KCD pit which were estimated at 8%.

Stopes mined in 2017 showed an “ore gain” of 4% or approximately 0.4 m thickness. “Ore gain” is material above the cut-off grade that was planned to be mined later in an adjacent stope, but due to overbreak has been mined early. It is assumed that continuous improvement of drill / blast processes will lead to a reduction in this overbreak. “Ore gain” has not been included in the estimation of Ore Reserves.

The Qualified Person has performed an independent verification of the block model tonnes and grade, and in their opinion, the process has been carried out to industry standards.

The Qualified Person is not aware of any environmental, legal, title, socioeconomic, marketing, mining, metallurgical, infrastructure, permitting, fiscal, or other relevant factors that could materially affect the Ore Reserve estimate.

18^th September 2018 Page 11

Kibali Gold Mine NI 43-101 Technical Report

1.8 Mining Method

Over the Life of Mine (LOM) of Kibali, a total of 64 Mt of ore at 4.16 g/t Au is expected to be produced over 19 years up to 2036. Ore supplied to the plant during this period, including stockpile changes, will be 66 Mt at an average grade of 4.09 g/t Au resulting in 7.6 Moz recovered at an average processing recovery of 89%.

The Kibali open pit operation will continue until 2026 and the underground until 2036.

A total of 43 Mt of ore will be mined from the underground operations with a further 22 Mt mined from open pits

In the Qualified Person’s opinion, the parameters used in the Mineral Resource to Ore Reserve conversion process are reasonable.

Open Pits

Open pit mining takes place in a number of satellite pits over approximately 14 km. Some of the pits are relatively shallow and have a short mine life of two years or less such as Pamao and Sessenge, whilst others are deeper and have a longer life of more than two years, such as Pakaka and Gorumbwa. There are six main open pit deposits, KCD, Pakaka, Pamao, Kombokolo, Sessenge, and Mengu Hill, located within an approximately 7 km radius.

The KCD pit will be the largest pit at 1.7 km N-S (approximately), 0.8 km E-W and 250 m deep. Mining has now been completed at the Mofu (2015), Mengu Hill, and Rhino (2016) pits, and at the first two pushbacks in the KCD pit (2016).    

As of 31^st December 2017, KCD pushback 3, Pakaka, and Kombokolo pits are the operational pits; Sessenge and Gorumbwa open pits are in the pre-production stage; however, mining of Pakaka is planned to be terminated in 2018, and the next pushback is not scheduled for production until 2020.

Open pit mining is conducted by contractor Kibali Mining Services (KMS), a local subsidiary of DTP Terrassement, using either free-dig or conventional drill, blast, load, and haul methods.

The mining equipment is jointly owned by a subsidiary of Randgold and the contractor’s parent, which also operates at Randgold’s Loulo-Gounkoto Mine in Mali and Tongon Mine in Côte d’Ivoire.

All the mineral deposits are characterised by the presence of a near-surface groundwater table with the potential for high groundwater inflows into the pits. The possible impacts of ingress of groundwater are investigated prior to mining and during the mining activities. Dewatering well systems are installed for all pits to lower the groundwater level prior to commencement of mining. A system of dewatering trenches are procedurally established prior to commencement of mining in each of the pits, preventing the inflow of any surface water to the active mining areas.

18^th September 2018 Page 12

Kibali Gold Mine NI 43-101 Technical Report

The upper levels of the open pits are usually in weathered material, which typically is free digging material. Once fresh (unweathered) rock is encountered, drilling and blasting is required. Emulsion explosives are supplied as a down-the-hole service by Orica.

Free digging in the upper levels uses 5 m high benches, with 10 m benches used for drilling and blasting operations. The 10 m benches containing ore are excavated in three flitches of equal height.

Opportunities exist to upgrade and convert the Inferred Mineral Resources within the current pits to Ore Reserves with drilling, but any Inferred Resources within pit designs are not reported as Ore Reserves. When the underground is at full capacity in 2019, a reduction in open pit production will be possible.

Under current Ore Reserves, a ramp down of open pit production is scheduled to begin from 2023. The open pit end of life is estimated at year 2026 based on current Ore Reserves. The addition of future open pit Reserves from additional exploration sites such as Kalimva have the potential to extend surface mining post 2026

Underground

The Kibali KCD underground mine is designed to extract the KCD deposit directly beneath the KCD open pit. A 50 m crown pillar separates the pit bottom from the top of the underground mine. The underground mine is a long hole stoping operation planned to produce at a rate of 3.6 M ore tonnes per year.

Development of the underground mine commenced in 2013. Stoping within the upper levels commenced in 2015, utilising the twin surface decline system for trucking of ore to surface. A vertical production shaft (751 m deep) is scheduled for full commissioning during 2018 following commissioning of the materials handling system completed at the end of 2017. Consequently, underground production is scheduled to peak at 320 kt per month by the fourth quarter of 2018 as annual underground production ramps up to achieve the 3.6 Mt design capacity.

A major pump station has been installed near the shaft bottom with redundant capacity in the pumps and pipelines to the surface.

In 2018, the production schedule will see the majority of ore being hoisted up the shaft, but throughout the underground life of mine (LOM) the decline to surface will be used to haul ore from the shallower zones and to supplement the shaft haulage.

A significant portion of the capital and access development for the mine is in place. To date, 27 km of capital development and 10 km of waste access development has been completed. The current LOM predicts a further 21 km of lateral capital development and 18 km of waste access development.

Ore from stopes is loaded (both by teleremote and conventional manual loaders) from the stopes into the eight ore passes via finger raises on the respective levels. This ore is then transferred by Autonomous load haul dumpers (LHDs) into two coarse ore bins and then into two primary crushers, followed by two fine ore bins and independent skip loadout conveyors near the shaft bottom.

18^th September 2018 Page 13

Kibali Gold Mine NI 43-101 Technical Report

The proposed mining methods are variants of long hole open stoping with cemented paste:

☐ Primary / Secondary long hole open stoping (primary 28% of Ore Reserve tonnes, secondary 40% of tonnes) is used in the wider zones, with 35 m interval heights where stopes are mined either as single lift or multiple (up to four) lifts, depending on stope geometry and the geotechnical stable span.

☐ Advancing face long hole open stoping (23% of Ore Reserve tonnes) is used in the deepest portion of the mine where the mineralisation has a shallower plunge (approximately. 20° to the NE), where stopes are mined with variable interval heights between 25 m and 35 m to optimise extraction.

☐ Longitudinal open stoping (9% of Ore Reserve tonnes) is used in narrow zones (< 15 m width) with variable interlevel heights between 20 m and 35 m.

No significant failures of the openings in the underground workings have occurred. The rock assessed for the rock mass model is ranked as good to very good.

The underground mining operations are currently operated by contractors (Byrnecut) and the contractor operation will continue until approximately mid to late 2018, when the changeover to The Kibali KCD underground mine has a scheduled production rate of 3.6 Mtpa for 10 years. The LOM predicts a long tail of declining production over a further nine years thereafter. The schedule is expected to be progressively optimised to extend the period of the 3.6 Mtpa production rate.

1.9 Mineral Processing

The Kibali gold processing plant comprises two largely independent processing circuits, the first one designed for oxide and transition ores and the second for sulphide refractory ore. However, both circuits are designed to process sulphide ore when the oxide and transition ore sources are no longer available. The flow sheet, depicted in Figure 1-1 comprises crushing, ball milling, classification, gravity recovery, a conventional CIL circuit, flash flotation, also conventional flotation, together producing a concentrate which goes to ultra-fine-grinding and a dedicated intensive cyanide leach. This process consists of well tested technology in the gold industry and is appropriate for Kibali’s style of mineralisation.

The extensive metallurgical testwork campaigns conducted for Kibali demonstrate two distinct behavioural patterns where some ore types exhibit free-milling characteristics suitable for gold extraction by a conventional carbon in leach (CIL) metallurgical process. Other ore types exhibit a degree of refractoriness, where straight cyanidation returns gold dissolutions considered to be too low for optimal plant operation. The reason for this refractoriness is due to the presence of occluded gold particles within sulphide minerals. It is found that a finer grind will expose a portion of this additional gold for leaching so that the recovery is enhanced to economically acceptable levels. Each process may be associated with the following standard recovery processes:

☐ Free-milling ores – conventional CIL circuit including gravity recovery.

☐ Partially refractory ores – includes a flotation circuit with ultra-fine-grinding (UFG) and dedicated intensive leaching of the concentrate generated.

18^th September 2018 Page 14

Kibali Gold Mine NI 43-101 Technical Report

Figure 1-1 Simplified Flowsheet of the Kibali Process Plant Depicting Two Discreet Streams

Figure 1-2 provides a historic testwork depiction of the variation of gold dissolution seen for the various Kibali ore types.

18^th September 2018 Page 15

Kibali Gold Mine NI 43-101 Technical Report

Figure 1-2 Initial Hole Composite Dissolutions (Blue Markers = Fresh, Green Markers = Transition, Red Markers = Oxide).

The current metallurgical recoveries expected at Kibali derived from both testwork and actual plant operational knowledge and used in the financial model can be found in Table 1-4.

Table 1-4 Summary of Average Recovery for All the Samples

Ore         Source	Weathering	Average Recovery     Primary Only     (%)	Average Recovery     Primary + Tail Leach (%)	Average Recovery     Oxide Process     (%)	Feasibility or Financial     Model Recovery (%)
KCD	Fresh_OP	86.4	89.2		86.1
	Fresh_UG	89	93.4		89.8
	Transition	66.6	91.3		90.1
	Oxide			89.1	85.8
Sessenge	Fresh	72.7	81.2		79.1
	Transition		80.3		75.9
	Oxide			90.4	90.3
Pakaka	Fresh	78.1	82.3		80.2
	Transition				81.3
	Oxide			96.9	88.7
Mengu Hill	Fresh	69.2	72.2		70.1
	Transition	84.4	89.9		89.3
	Oxide			92.6	89.3
Kombokolo	Fresh	70.3	75.2		73.1
	Transition	78.9	95.3		95.9
	Oxide			96.4	95.6
Pamao	Fresh	74.5	85.5		83.5
	Transition				85
	Oxide			95.8	90.9

Within the existing Kibali process plant, ore is blended using both KCD underground ore plus ore sourced from satellite open pits at Kibali. The process plant has demonstrated improvements in throughput capability, performing beyond design capacity of 7.2 Mtpa as depicted in Figure 1-3, at reasonably consistent recovery performance (Figure 1-4). The variation seen in the recovery performance is partially explained by the ore types being treated, in addition to there being a distinct overall improvement trend with time.

18^th September 2018 Page 16

Kibali Gold Mine NI 43-101 Technical Report

Figure 1-3 Graph Indicating Kibali Throughput Performance Against Design of 7.2Mtpa

Figure 1-4 Kibali Plant Process Recovery for Year 2017

18^th September 2018 Page 17

Kibali Gold Mine NI 43-101 Technical Report

Actual plant operating costs for 2016 and 2017 are denoted in Table 1-5, in line with budget forecast.

Table 1-5 Actual Process and Plant Engineering Operating Costs for 2016 and 2017

Cost	Units	2016 Actual	2017 Actual
Fixed Cost
Consultants	$’000	203	347
Contractors - Assays	$’000	1544	1,588
Contractors - Oxygen	$’000	1735	1,634
Equipment Hire	$’000	3299	2,852
General Costs	$’000	6655	6,903
Gold Refining	$’000	3233	3,917
Labour	$’000	4896	5,734
Stores - Other	$’000	391	1
Total Fixed	$’000	21,956	22,976
Tonnes Processed	kt	7296	7,619
Total Fixed	$/t	3.01	3.02
Variable Costs
Power	$/t	3.90	4.49
Reagents - Cyanide	$/t	3.10	2.80
Reagents - Lime	$/t	1.20	0.60
Good Issues -Caustic Soda	$/t	0.67	0.58
Good Issues -Activated Carbon	$/t	0.16	0.09
Reagents - Other	$/t	2.01	2.02
Stores - Grinding Media	$/t	0.81	0.96
Stores - Liners	$/t	0.41	0.43
Stores -Screens and Panels	$/t	0.01	0.10
Total Variable	$/t	12.27	12.07
Total $/ t	$/t	15.28	15.09
Plant Engineering	$/t	3.79	3.62
Combined Plant & Engineering	$/t	19.07	18.71

LOM processing costs have been budgeted at $17.24/t (which included plant engineering cost). The actual costs for 2017 were $18.71/t, with the key improvements over the LOM being the two hydropower stations that have been brought online in 2018 and 2019, which will drop power cost by approximately $1.30/t. Further to this, cyanide consumption has been optimised to a level of $2.50/t for the LOM (2017 levels being $2.80/t) as the plant operations are stabilised.

1.10 Project Infrastructure

Infrastructure in the DRC is generally poor as a result of limited investment in maintenance, upgrades and extensions of the road networks established during colonial times.

The Kibali site is located 185 km from Arua on the border with Uganda and all transport links take place through Uganda from Kenya, Sudan, and Tanzania. Most supplies come from Mombasa (1,800 km); however, Dar es Salaam (1,950 km) and Port Sudan provide alternative ports. The routes to the border with DRC are paved. The arterial road between Arua and the site is unpaved but has been upgraded and serves as the main material access route to site for the operations. Local

18^th September 2018 Page 18

Kibali Gold Mine NI 43-101 Technical Report

roads are in a very poor states of repair. Supplies typically require two weeks to arrive from Mombasa.

Internal roads provide access to various infrastructure areas, including roads to the TSF, Explosives Storage, Land Fill Site, Mine Villages (Senior and Junior), Central Mine Offices, Shaft Collar Area, Open Pit Mining Central Operations Area, general mining operations areas, new exploration areas, various water boreholes, and overhead line routes.

Daily flights with international air carriers are available from Entebbe. Charter flights between Entebbe and the (unsealed) airstrip at the mine are available, when required. Surface water run-off is high, due to high intensity rainfall events and an undulating landscape.

The primary source of raw water is rain and spring water catchments with a dam top-up from a borehole system and final backup from the Kibali River. Raw water is collected and stored in the raw water dam (RWD), which has a storage capacity of 9,500 m³. A system of bund walls and dewatering trenches has been established. The network of drainage channels is used to discharge water intercepted by the perimeter drains to the Kibali River via a series of settling ponds.

The processing plant is supplied by return water from the tailings storage facilities (TSFs), thickener overflow, and storm water. The operational camp has an independent water purification plant and storage facility.

Two TSFs exist at Kibali; one for the CIL tails and one for the Flotation tails. The tailings from all ore and concentrate that have been processed using cyanide leaching are pumped to the Cyanide Tailings Storage Facility (CTSF) which is an above ground storage facility lined with high-density polyethylene (HDPE).

The benign tailings from the flotation circuit treating the sulphide ore is pumped to the Flotation Tailings Storage Facility (FTSF) which is a valley fill dam formed by an embankment across a nearby valley. This does not have an installed HDPE lining.

Approximately half of the sulphide tailings generated will be used to produce paste backfill for the stoping operations. A paste fill plant filters the sulphide tailings, which are mixed with cement to form a paste fill that is delivered to the underground via a distribution pipe network from the surface.

Kibali is totally dependent on its own power generation facilities for the supply of electrical power. There are three separate thermal power stations that each have twelve 1500 kVA Cat diesel generators. For purposes of reducing Kibali’s reliance on thermal generation and reducing the mine operating costs, several feasibility studies were undertaken for the justification of regional hydropower installations. To date two such plants have been installed with a third nearing completion. These have had a marked effect on reducing the unit power operating cost on a $/kWh basis.

The hydrostation installations are Nzoro 2 (four 5.5 MW turbines), Ambarau (two 5.3 MW turbines), and Azambi (two 5.3 MW turbines). Nzoro 1 is pre-existing legacy hydrostation built in the 1930s and has a capacity of less than 1 MW. It was refurbished by Kibali Goldmines as part an agreement

18^th September 2018 Page 19

Kibali Gold Mine NI 43-101 Technical Report

with SOKIMO, such that the produced power is dedicated to providing power to the local communities.

Nzoro 2 was commissioned successfully, being implemented at the same time as the main Kibali mine development project. The Ambarau hydropower generation plant was completed subsequent to the commissioning of Nzoro 2 in January 2017. The Azambi hydropower generation plant is scheduled for completion in Q3 2018.

Therefore, the system has a potential capacity of 44 MW of hydropower (at peak) and 32 MW of thermal Gen-sets. Actual hydro generation capacity is season dependent:

☐ Maximum Capacity (32 + 44) MW

☐ Minimum Capacity (32 + 10) MW.

The load demand of the mine is not constant, and the average power consumption will be approximately 40 MW.

1.11 Market Studies

Financial evaluation of all Ore Reserves uses a gold price of $1,000/oz and with the exception of Ore Reserves for the KCD pit which uses a $1,100/oz gold price optimised pit design. All other open pit Ore Reserves are estimated within pit designs which are based on a gold price of $1,000/oz. This is in line with Randgold’s corporate guidelines. Gold price sensitivities were run for all the pits and the decision on a higher price for the KCD is discussed in more detail in Section 15.

Financial evaluation and cut-off grade calculation for the Kibali underground Ore Reserves has been based on a gold price of $1,000/oz. This same value has been used for all previous Kibali underground Ore Reserves estimate from 31^st December 2011 onwards.

Royalties payable to the DRC government remain unchanged from completion of the feasibility in 2012. A total royalty payable to the DRC government of 3.5% of gold revenue inclusive of 1% shipment fees was used for the open pit Ore Reserve estimate.

Kibali currently pays income tax at a rate of 30% to the DRC government. Due to accelerated depreciation charged on capital expenditure, a tax shield has been built up, meaning taxation payment will only commence in 2024.

Gold doré produced at the mine site is shipped from site under secured conditions and sold under agreement to Rand Refinery in South Africa. Under the agreement, Kibali Goldmines receives the ruling gold price on the day after dispatch, less refining and freight costs, for the gold content of the doré gold. Kibali Goldmines has an agreement to sell all gold production to only one customer. The “customer” is chosen periodically on a tender basis from a selected pool of accredited refineries and international banks to ensure competitive refining and freight costs. Gold mines do not compete to sell their product given that the price is not controlled by the producers.

18^th September 2018 Page 20

Kibali Gold Mine NI 43-101 Technical Report

1.12 Environmental, Permitting and Social Considerations

An independent Environmental and Social Impact Assessment (ESIA) for the Kibali mine was completed as part of the Kibali Goldmines Feasibility Study completed in December 2012. Subsequent, ESIAs for various Project extensions and new elements were completed over the next four years and these were consolidated in 2016. An Environmental Adjustment Plan (EAP) has been approved by the Direction de Protection de l’Environnement Minier (DPEM) with the purpose of describing any measures that have been or will be taken for the purpose of the protection of the environment. An environmental management plan is in place, and the Kibali operations are ISO 14001:2015 compliant and independently audited to continuously improve environmental management. The site is also audited against the requirements of the International Cyanide Management Code.

Waste rock is generated and disposed of close to the open pits. The waste rock characterisation assessment returned a negative acid generating status. Waste rock is used to build various infrastructural platforms on site, while the remainder is stockpiled on surface or deposited in stopes as backfill. The waste rock has been demonstrated to have moderate to high acid neutralising capacity for the majority of lithologies tested.

Tailings are generated from the plant and disposed of in two separate tailings storage facilities, the flotation TSF (FTSF) and concentrate TSF (CTSF), which consists of the CTSF1 and CTSF2. The CTSF is lined and contains materials which are acid producing and which also contain cyanide residues and materials with a higher arsenic content. The CTSF is due to be extended and authorisation for this will be applied for in 2018. A portion of tailings is currently used for paste backfill in the underground KCD operation, which will be at full capacity in 2018.

Routine environmental monitoring takes place across the site, including dust deposition, noise, arsenic, and weak-acid dissociated (WAD) cyanide sampling, TSF seepage water and tails streams as well as sample collection of drinking water, ground water, surface water and the TSF borehole water.

Environmental incidents are noted in a register which forms part of the Environmental Management System (EMS); the causes and responses are identified, and once completed, the incident is closed out. There were no reported major incidents in 2017.

A comprehensive water balance model has recently been developed for the site, which models flows, inputs and losses across the complex site, including the open pits and underground workings, plant, TSF, water management structures, offices, camp, and treatment facilities.

The original vegetation of the Project area has been largely transformed through human activity. Three plant species were recorded within the Project area which are considered to be of conservation significance by the IUCN. No faunal species of international conservation significance were identified during the surveys. Despite human pressure, both the gallery forest and the moist savannah are in fairly good condition and are home to several habitat specific species.

18^th September 2018 Page 21

Kibali Gold Mine NI 43-101 Technical Report

Biodiversity monitoring is ongoing, such as the use of camera traps to detect fauna within the concession. The Biodiversity Management Plan (BMP) is being updated to reflect additional information on the biodiversity which has been collected. The mine site lies around 65 km south of the Garamba National Park, which lies on the border with South Sudan. A partnership with the Park has been established to support the Park’s goals. This partnership provides a wider strategic support for game protection from poachers from the north, and connections with local enforcement networks.

Mine closure costs are updated each year, with increases or decreases in disturbed areas noted and costed; the current cost for rehabilitation and closure of the mine according to the calculation model is $32 M.

The mine is a significant employer to members of the local communities. The mining operations contribute to extended life-of-mine, employment of local Congolese and the growth of the DRC economy. Kibali Goldmines policy is to promote nationals to manage the Project. The policy of promoting local employment also extends to its contractors. Overall local employees’ number 4,917 out of a total workforce of 5,377 employer and contractor employees. Local procurement is also promoted and is a contractual requirement for contractors.

Due to the construction of the Project, it was necessary to resettle approximately 17,000 people from the immediate Project area, referred to as the Exclusion Zone. Furthermore, the Project displaced around 134 items of community infrastructure, including 13 communal agricultural projects, five communal business/commercial facilities, 12 education facilities, 19 health facilities, nine recreational/community facilities, 39 religious facilities and 41 water sources. This approved resettlement plan was carried out in 2012 and 2013.

Gorumbwa Resettlement Action Plan (RAP) was initiated in 2016 to allow for future mining. So far, 1,329 households have received cash compensation, 1,397 houses are in various stages of construction, 144 nearing completion, 706 are inhabited and 42 completed, awaiting physical movement.

Stakeholder engagement activities, community development projects and local economic development initiatives contribute to the maintenance and strengthening of Kibali Goldmines Social License to Operate (SLTO). A grievance mechanism is in place, and all registered grievances in 2017 were successfully resolved.

Artisanal mining remains a concern in the Kibali Permit area and the mine is working with provincial authorities to eliminate artisanal and small-scale mining (ASM) within the Permit.

The Qualified Persons consider the extent of all environmental liabilities to which the property is subject to have been appropriately met.

1.13 Capital Costs

Kibali is an on-going combined open pit and underground mining operation with the necessary facilities, equipment, and manpower in place to produce gold.

18^th September 2018 Page 22

Kibali Gold Mine NI 43-101 Technical Report

The open pit and underground LOM and capital and operating cost estimates have been completed in sufficient detail to be satisfied that economic extraction of the Proved and Probable Ore Reserves is justified.

The underground development and the shaft completion ($132.2 M), along with the completion of the second and start of the third hydropower stations, were the key capital projects for 2017. Additional capital expenditure was incurred on the expansion of the Ultra-Fine Grind capacity, plus deferred stripping at both Kombokolo and Pakaka satellite pits, the Gorumbwa Resettlement and rebuild costs to the open pit mining fleet.

The majority of the capital cost estimates contained in this report are based on quantities generated from the open pit and underground development requirements.

Capital expenditure over the remaining LOM is estimated to be $370.6 M, made up from the allocation of costs as summarised in Table 1-6.

Table 1-6 LOM Capital Expenditure

Description	Value         ($ ‘M)
Construction and Project Capitals	22
Ongoing Capital	99
Underground Capital Development and Drilling	203
Pre-Production Capitalised	9.3
Exploration Capitalised	5.4
Rehabilitation/Mine Closure	32
Total LOM Capital Expenditure	371

1.14 Operating Costs

Kibali maintains detailed operating cost records that provide a sound basis for estimating future operating costs.

Costs used for the open pit optimisations were derived from the Mining Contractor’s pricing of the open pit LOM schedule. Owners cost were also added for underground operations as of the third quarter of 2018 in preparation for the move to Owner operations underground.

Labour costs for national employees were based on actual costs. Local labour laws regarding hours of work, employment conditions were also considered and overtime costs included.

During 2017 costs for processing and general and administration (G&A) were updated based on actuals adjusted with the latest forward estimates, production profiles and manning levels.

Customs duties, taxes, charges and logistical costs have been included.

18^th September 2018 Page 23

Kibali Gold Mine NI 43-101 Technical Report

Unit costs used to estimate LOM operating costs are summarised in Table 1-7. The annual fluctuation in production levels is relatively low, such that the effect of fixed versus variable expenses is minimised.

For the underground mine, operating costs have been derived from 2017 actual costs for Kibali. LOM costs have been adjusted to reflect operational changes including the move from contract mining to Owner operations during 2018.

Table 1-7 LOM Unit Operating Costs

Activity	Units	Value
Open Pit Mining - Kibali	$/t mined	3.27
Open Pit Mining - Kibali	$/ore tonne mined	21.62
Underground Mining	$/t mined	34.46
Underground Mining	$/ore tonne mined	35.88
Stockpile Movement	$/t milled	0.30
Processing	$/t milled	17.20
G&A	$/t milled	7.78
Mining Total	$/t milled	30.40
Total LOM Net OPEX	$/t milled	55.58

The LOM has been prepared on the basis that the underground mining activities will transition to an Owner operated mine during 2018. It is assumed that current contract prices will remain unchanged for mining activities performed by a contractor such as open pit mining and the underground development and production.

Cost inputs have been priced in real Q4 2017 dollars, without any allowance for inflation or consideration to changes in foreign exchange rates.

The Qualified Persons are satisfied that the open pit LOM and cost estimates have been completed in sufficient detail to justify the economic extraction of the open pit Proved and Probable Ore Reserves.

The Qualified Persons are satisfied that the underground LOM and cost estimates have been completed in sufficient detail to justify the economic extraction of the underground Proved and Probable Ore Reserves.

1.15 Economic Analysis

This section is not required as the property is currently in production, Kibali Goldmines is a producing issuer, and there is no material expansion of current production. The Qualified Person has verified the economic viability of the Ore Reserves via cash flow modelling, using the inputs discussed in this report.

18^th September 2018 Page 24

Kibali Gold Mine NI 43-101 Technical Report

1.16 Interpretation and Conclusions

Geology and Mineral Resources

Kibali Goldmines has documented standard procedures for the drilling, logging, and sampling processes, which meet industry standards. The geological and mineralisation modelling at Kibali is based on visibly identifiable geological contacts, which ensure a geologically robust interpretation can be developed.

Kibali has a quality control program in place to ensure the accuracy and precision of the assay results from the analytical laboratory. Checks conducted on the quality control database indicated that the results are of acceptable precision and accuracy for use in Mineral Resource estimation.

Geological models and subsequent Mineral Resource estimations have evolved and improved with each successive model update from added data within both open pit and underground. Significant grade control drill programs, and mapping of exposures in mine developments have been completed to increase the confidence in the resulting Mineral Resources and Ore Reserves. This was demonstrated in 2017 as this was the first time that Proved Ore Reserves have been disclosed for the underground mine.

In the Qualified Person’s opinion, the Kibali Mineral Resources top capping, domaining and estimation approach are appropriate, using industry accepted methods. Furthermore, the constraint of underground Mineral Resource reporting to use optimised mineable stope shapes has been deemed to reflect good to world best practice by external project audits. The Qualified Person considers the Mineral Resources at Kibali are appropriately estimated and classified.

The Qualified Person is not aware of any environmental, permitting, legal, title, socioeconomic, marketing, metallurgical, fiscal, or other relevant factors, which could materially affect the Mineral Resource estimate.

The strategic focus of Kibali exploration is to prioritise additions of resources and reserves at open pit satellite projects to extend the life of open pit operations. Additionally, underground resource definition down plunge extension drilling is currently focussed on the target areas above the base level of the shaft thereby decreasing potential development requirements to access such areas for production.

Mining and Ore Reserves

The open pit mining operations at Kibali consists of multiple open pits. The open pits are being operated by a mining contractor and a down-the-hole blasting service will be provided by an appropriate blasting contractor. Opportunities exist with the Inferred Mineral Resource within the current pits that can be upgraded and converted to Ore Reserve with drilling. A reduction of open pit production is scheduled from 2023 and the end of current open pit mine life is estimated at year 2026 based on current Ore Reserves.

18^th September 2018 Page 25

Kibali Gold Mine NI 43-101 Technical Report

The Kibali KCD underground mine is designed to extract the KCD deposit directly beneath the KCD pit. A 50 m crown pillar separates the pit bottom from the top of the underground mine. The underground mine is a long hole stoping operation planned to produce at a rate of 3.6 Mtpa for 10 years. The majority of the underground mine infrastructure is already in place. A vertical production shaft is scheduled for full commissioning during 2018 following commissioning of the materials handling system. In 2018, the production will move to the majority of ore being hoisted up the shaft, however, throughout the underground LOM the decline to surface will be used to haul ore from some of the shallower zones and to supplement the shaft haulage.

The LOM has a long tail of declining production over a further nine years. The schedule will be progressively optimised as underground exploration of down plunge extensions progresses to extend the period of 3.6 Mtpa production rate.

Kibali Goldmines, as the owner operator of the Project has significant experience in other mining operations within Africa and these production rates, modification factors, and costs are benchmarked against other African operations to ensure they are suitable, taking into account the increased relative cost of fuel and labour within the DRC.

The current Ore Reserves for Kibali support a total mine life of 15 years at near full mill capacity, nine years of open pit operations, and 15 years of underground mining. LOM gold production averages approximately 542 koz per year.

The Qualified Person considers the modelled recoveries for all ore sources and process plant combined process and engineering unit costs, used within the Mineral Resource and Ore Reserve process to be acceptable.

The Qualified Person is not aware of any environmental, legal, title, socioeconomic, marketing, mining, metallurgical, infrastructure, permitting, fiscal, or other relevant factors that could materially affect the Ore Reserve estimate.

Processing

Extensive metallurgical testwork campaigns have been completed across all mineral deposits in Kibali that form part of the declared Ore Reserve. These have consistently demonstrated two distinct behavioural patterns the first of which exhibits free-milling characteristics suitable for gold extraction by a conventional CIL metallurgical process and the second of which exhibits a degree of refractoriness, where straight cyanidation returns gold dissolutions considered to be too low for optimal plant operation due to the presence of occluded gold particles within sulphide minerals. It has been demonstrated that a finer grind will expose a portion of this additional gold for leaching so that the recovery is enhanced to economically acceptable levels.

The Kibali process plant operational risks are materially reduced as a function of the two separate process streams and independent milling circuits. The process plant has demonstrated excellent improvements in throughput capability, even performing beyond design capacity at 7.2 Mtpa at consistent recovery performance.

18^th September 2018 Page 26

Kibali Gold Mine NI 43-101 Technical Report

The ore feed plan is blended using both KCD underground ore plus ore sourced from satellite open pits at Kibali in order to provide a stable feed grade blend. The Kibali feed plan utilises geometallurgical models that estimate the arsenic content within potentially arsenic bearing mineral deposits such that any ore with high arsenic contents is stockpiled separately and blended into the CIL process route to ensure that discharge is directed to the lined CTSF and discharge levels are below the environmental requirements.

The Qualified Person considers the modelled recoveries for all ore sources and the process plant and engineering unit costs applied to the Mineral Resource and Ore Reserve process to be acceptable.

Environment and Social

Kibali has a maturing environmental and social management plan and an accredited ISO14001:2015 Environmental Management System (EMS) in place which addresses current operational needs and can readily be adapted to meet future activities. Mine closure costs are reviewed and revised annually in line with good international industry practice.

All permits are in place and an Environmental Adjustment Plan has been approved by the DPEM.

The mine prioritises local employment and in 2017 the workforce was made up of 83% contractors, of whom 92% were nationals, and 17% employees, of whom 88% were nationals. Overall local employees number 4,917 (92%), out of a total of 5,377 employer and contractor posts.

Stakeholder engagement is ongoing, and all senior management are involved in regular meetings with the community.

Two significant resettlement campaigns have taken place, one in 2012/2013 and one in 2016/2017. Ongoing monitoring of affected households to ensure that their livelihoods, often previously based on artisanal mining, are not adversely affected by the resettlement, will be ongoing. Economic displacement has also been significant across the area.

Artisanal mining remains a concern in the Kibali Permit area and the mine is working with provincial authorities to eliminate ASM within the Permit.

Kibali Goldmines continues to invest in community development initiatives, focussing on potable water supplies, primary school education, health care education, investment in medical clinics and local economic development projects.

The mine is a significant employer to members of the local communities. The mining operations contribute to extended life-of-mine, employment of local Congolese and the growth of the DRC economy. Kibali Goldmines policy is to promote nationals to manage the project.

The Qualified Person considers the extent of all environmental liabilities, to which the property is subject, to have been appropriately met.

18^th September 2018 Page 27

Kibali Gold Mine NI 43-101 Technical Report

Ownership and DRC Mining Code

The Kibali operation conforms to the DRC mining code and regulations. The next renewal date for the Permits are on 5^th November 2029 and 6^th March 2030 and the current life of mine plan for the Kibali Ore Reserves extends beyond these dates.

The DRC mining code (2002) includes provision for renewal of all exploitation licences for a successive period of 15 years, providing the holder has not breached the licence obligations of license fee an annual surface rights fees payment and upholds all environmental standards set out in the exploitation Permit. Additionally, the Permit holder should provide the appropriate government departments with a monthly mining activity report and quarterly exploration reports.

The DRC Mining Code (2002) and Regulations have been amended with an updated Mining Code which came into force on 9th March 2018 (DRC Mining Code(2018)) and the related amended Mining Regulations which came into force on 8th June 2018.

Going forward the DRC Mining Code (2018) envisages a stability period for the tax, customs and exchange control regime of five years from the date on which the DRC Mining Code (2018) came into force and further provides that a number of the taxes shall be applied in accordance with the applicable substantive law.

The QP notes that the mining code is in the process of transition and the current proposed changes do not have an impact on the stated Mineral Resource or Ore Reserves at the gold prices used for Ore Reserve ($1,000/oz), Mineral Resource ($1,500/oz), or the current gold prices of $1,200/oz (August 2018).

All Permit fees, surface rights fees, and taxes relating to Kibali’s exploitation rights have been paid to date and reporting requirements conformed to, accordingly the concession is in good standing. At the time of compiling this report, the Qualified Person is not aware of any risks that could result in the loss of ownership of the deposits or loss of the Permits, in part or in whole.

Infrastructure

Kibali is a mature operation that has all necessary support infrastructure already in place.

For purposes of reducing Kibali’s reliance on thermal generation and reducing the mine operating costs, three hydropower stations with a potential capacity of 44 MW of hydropower (at peak) and 32 MW of thermal Gen-sets. The average annual power consumption of the mine operation is approximately 40 MW, which the three hydropower stations are designed to supply approximately 80% of the mine demand taking into account fluctuation during the rainy seasons. Mine operating costs will be expected to be reduced in line with the introduction of hydropower station.

1.17 Risks

Kibali Goldmines has undertaken analysis of the Project risks. Table 1-8 summarises the Project risks and the Qualified Persons assessment of the risk degrees and consequences, as well as

18^th September 2018 Page 28

Kibali Gold Mine NI 43-101 Technical Report

ongoing/required mitigation measures. The Qualified Persons note that the degree of risk refers to our subjective assessment as to how the identified risk could affect the achievement of the Project objectives.

Kibali has been in production for five years and is a mature operation

In the Qualified Persons opinion, there are no significant risks and uncertainties that could reasonably be expected to affect the reliability or confidence in the exploration information, Mineral Resource or Ore Reserve estimates.

Risk Analysis Definitions

The following definitions have been employed by the Qualified Persons in assigning risk factors to the various aspects and components of the Project:

☐ Low – Risks that are considered to be average or typical for a deposit of this nature and could have a relatively insignificant impact on the economics. These generally can be mitigated by normal management processes combined with minor cost adjustments or schedule allowances.

☐ Minor – Risks that have a measurable impact on the quality of the estimate but not sufficient to have a significant impact on the economics. These generally can be mitigated by normal management processes combined with minor cost adjustments or schedule allowances.

☐ Moderate – Risks that are considered to be average or typical for a deposit of this nature but could have a more significant impact on the economics. These risks are generally recognisable and, through good planning and technical practices, can be minimised so that the impact on the deposit or its economics is manageable.

☐ Major – Risks that have a definite, significant, and measurable impact on the economics. This may include basic errors or substandard quality in the basis of estimate studies or Project definition. These risks can be mitigated through further study and expenditure that may be significant. Included in this category may be environmental/social non-compliance, particularly in regard to Equator Principles and International Finance Corporation (IFC) Performance Standards.

☐ High – Risks that are largely uncontrollable, unpredictable, unusual, or are considered not to be typical for a deposit of a particular type. Good technical practices and quality planning are no guarantee of successful exploitation. These risks can have a major impact on the economics of the deposit including significant disruption of schedule, significant cost increases, and degradation of physical performance. These risks cannot likely be mitigated through further study or expenditure.

In addition to assigning risk factors, the Qualified Persons provided opinion on the probability of the risk occurring during the LOM. The following definitions have been employed by the Qualified Persons in assigning probability of the risk occurring:

☐ Rare – The risk is very unlikely to occur during the Project life.

☐ Unlikely – The risk is more likely not to occur than occur during the Project life.

☐ Possible – There is an increased probability that the risk will occur during the Project life.

18^th September 2018 Page 29

Kibali Gold Mine NI 43-101 Technical Report

☐ Likely – The risk is likely to occur during the Project life.

☐ Almost Certain – The risk is expected to occur during the Project life.

Risk Analysis Table

Table 1-8 Kibali Risk Analysis

Issue	Likelihood	Consequence Rating	Risk Rating	Mitigation
Geology and Mineral Resources – Confidence in Mineral Resource Models	Unlikely	Minor	Low	Additional scheduled infill drilling. Resource model updated on a regular basis using production reconciliation results.
Mining and Ore Reserves – Open Pit Slope Stability	Unlikely	Moderate	Minor	Continued in-pit monitoring, geotechnical drilling, instrumentation, and continued updating of geotechnical and hydrological models.
Mining and Ore Reserves – Underground Recovery and Dilution	Possible	Moderate	Low	Change in blasting practices to increase recovery and reduce dilution.
Processing - Water Dilution in CIL	Possible	Moderate	Medium	Several campaigns already successfully completed Plant modifications already installed to reduce water dilution at harvest screen and treatment of elution barrens
Processing Viscosity Drop in CIL Circuit Leading to carbon settlement	Possible	Moderate	Medium	Extensive trials performed. Viscosity modifiers or ensuring operation at acceptable slurry densities. Possible to increase the viscosity within the CIL by introducing oxides to supplement other ore or concentrate flows.
Environmental – Groundwater contamination (As) – Tailings failure	Possible	Major	Low	Manage As levels through feed profile. All high As feed reports to lined tailings facility. Continuing monitoring and external or third-party audits.
Social – Social License to Operate	Possible	Moderate	Moderate	Dedicated community engagement by company social and sustainability department.
Country & Political – Security – Governmental	Possible	Major	Moderate	Dedicated government liaison team in Kinshasa. Government participation/ownership.
Capital and Operating Costs	Unlikely	Moderate	Low	Continue to track actual costs and LOM forecast costs, including considerations for inflation and foreign exchange. Switching to Owner/Operator for underground mining in 2018
Fiscal Stability	Possible	Moderate	Moderate	Dedicated government liaison team in Kinshasa Government participation/ownership

18^th September 2018 Page 30

Kibali Gold Mine NI 43-101 Technical Report

1.18 Recommendations

The Qualified Persons make the following recommendations:

☐ A digital logging and data capture system to minimise manual data capture should be implemented.

☐ The 2018 mining sequence in the 5102 and lower 5101 zones has been modified. A full geotechnical review and model of mining induced stresses should be completed and if required the 2018 mining sequence adjusted.

☐ The following KPIs should be introduced for review at the time of the six-monthly update of the LOM plan to provide a quantifiable measure of the confidence in the change of plan:

o Number of continuous months in the LOM plan where 90% of scheduled stope production has had all resource grade control diamond drilling completed.

o Number of continuous months in the LOM plan where 90% of cross-cut and ore drive development has had all grade control diamond drilling completed.

☐ A professional development programme should be implemented aimed at developing suitably qualified resource geologists, mining engineers, and metallurgist to Qualified Person status.

☐ The option to further optimise the leach through configuration of concentrate in both the pre-oxidation and the main CIL circuits, thereby improving process recoveries should be further considered.

☐ To further decrease the mine’s reliance on thermal power and potentially reduce operating costs, further sites suitable for hydropower generation in the Kibali region should be identified and subjected to independent feasibility studies.

☐ An ASM cessation strategy should be agreed with the Haut Uélé governor so that the local community and local chiefs are sensitised to the importance of limiting ASM activities within the government identified ‘corridors’.

☐ Active and direct engagement with the government regarding the impact of new proposed mining code

18^th September 2018 Page 31

Kibali Gold Mine NI 43-101 Technical Report

2 Introduction

2.1 Introduction

The purpose of this report is to support the public disclosure of the 2017 Mineral Resource and Ore Reserve estimates at the Kibali Gold Mine (Kibali, the Mine or the Project) located in the Democratic Republic of the Congo (DRC) as of 31^st December 2017. This Technical Report conforms to National Instrument 43-101 - Standards of Disclosure for Mineral Projects (NI 43-101). All currency in this report is US dollars ($) unless otherwise noted.

Kibali Goldmines SA (Kibali Goldmines) is an exploration and mining company which is currently owned 70% by Moto Gold Mines Limited (Moto), and 20% by Kibali Jersey Ltd. Both Moto and Kibali Jersey Ltd. are joint ventures, owned 50% by Randgold Resources Limited (Randgold) and 50% AngloGold Ashanti Limited (AngloGold Ashanti). This equates to an overall interest in Kibali Goldmines of 45% for Randgold and 45% for AngloGold Ashanti. The remaining 10% of Kibali Goldmines SA is owned by Congolese parastatal Société Miniere de Kilo-Moto SA UNISARL (SOKIMO) with the shareholding held by the Minister of Portfolio of DRC (MoP).

Randgold is the operator at Kibali for both exploration and mining. In addition, Randgold is developing and operating gold mines in West and East Africa. The most notable of these are the following:

o Morila Gold Mine in Mali,

o Loulo Gold Mine in Mali,

o Gounkoto Gold Mine in Mali,

o Tongon Gold Mine in the Ivory Coast, and

o Massawa Exploration Project in Senegal.

The Project is an operating mine consisting of the Kibali Karagba-Chauffeur-Durba (KCD) underground mine, the KCD open pit, satellite deposits, a processing plant (7.2 Mtpa capacity), three hydropower stations, together with other associated mine operation and regional exploration infrastructure. The plant produces gold doré bars.

Total mine production from underground and open pits in 2017 was 7.6 Mt at a head grade of 2.9 g/t Au for a total of 596 koz gold (83.4% recovery). The shaft infrastructure was completed and commissioned in 2017. Underground mining continued to ramp-up during the year, which accounted for 1.8 Mt of this production.

Total production since mining commenced to year end 2017 is 161 Mt (30 Mt ore) for 2.4 Moz of gold (82.3% recovery).

The Mineral Resource and Ore Reserve estimates have been prepared according to the guidelines Australasian Code for Reporting of Exploration Results, Mineral Resources and Ore Reserves standards and guidelines published and maintained by the Joint Ore Reserves Committee of the

18^th September 2018 Page 32

Kibali Gold Mine NI 43-101 Technical Report

Australasian Institute of Mining and Metallurgy, Australian Institute of Geoscientists and Minerals Council of Australia (the JORC (2012) Code). Randgold has reconciled the Mineral Resources and Ore Reserves to Canadian Institute of Mining, Metallurgy and Petroleum (CIM) 2014 Definition Standards for Mineral Resources and Mineral Reserves dated May 10, 2014 (CIM (2014) Standards) as incorporated with NI 43-101 and there are no material differences.

2.2 Effective Date

The effective date of this report is 31^st December 2017.

2.3 Sources of Information

This Technical Report was prepared by Randgold on behalf of Kibali Goldmines and incorporates the work of Optiro Pty Ltd (Optiro) and Digby Wells and Associates Pty Ltd. (Digby Wells). The dates of personal inspections by the QPs are provided in Section 29 of this Technical Report.

The QPs and their responsibilities for this Technical Report are listed in Section 29 Certificate of Qualified Person and noted below:

☐ Mr. Simon Bottoms, CGeol, MGeol, FGS, MAusIMM (Randgold), is responsible for the preparation of sections 4 to 12, 14, and 23 of this Technical Report and shares responsibility with my co-authors for sections 1, 2, 3, 24, 25, 26, and 27.

☐ Mr. Rodney B. Quick MSc, Pr. Sci.Nat (Randgold), is responsible for the preparation of sections 19, and 22 of this Technical Report and shares responsibility with my co-authors for sections 1, 2, 3, 21, 24, 25, 26, and 27.

☐ Mr. Richard Quarmby, BSc, Pr Eng & C Eng, MSAIChE, MIoMMM, MBA (Randgold), responsible for the preparation of sections 13, 17 and 18 of this Technical Report and shares responsibility with my co-authors for sections 1, 2, 3, 21, 25, 26, and 27.

☐ Mr. Andrew Law, HND(MMin), MBA, FAusIMM (CP), FIQA, MAICD, AFAIM (Optiro), is responsible for the preparation of sections 15, and 16 of this Technical Report and shares responsibility with my co-authors for sections 1, 2, 3, 25, 26, and 27.

☐ Mr. Graham E. Trusler, Msc, Pr Eng, MIChE, MSAIChE (Digby Wells), is responsible for the preparation of section 20 of this Technical Report and shares responsibility with my co-authors for sections 1, 2, 3, 25, 26, and 27.

☐ The documentation reviewed, and other sources of information, are listed at the end of this report in Section 27 References.

18^th September 2018 Page 33

Kibali Gold Mine NI 43-101 Technical Report

2.4 List of Abbreviations

Units of measurement used in this report conform to the metric system. All currency in this report is US dollars ($) unless otherwise noted.

AA	Atomic Absorption
AARL	Anglo American Research Laboratory
ACSA	Albite-Carbonate-Silica Alteration
ALS	ALS Laboratories
AMTEC	AMTEC Laboratories
AMTEL	AMTEL Laboratory, Canada
ANSUL	Fire Suppression Supply Company
ARD	Acid Rock Drainage
ASM	Artisanal and Small-Scale Mining
BIF	Banded Ironstone Formation
BM	Block Model
BMP	Biodiversity Management Plan
BRT	Bottle Roll Test
CA	Confidentiality Agreement
CHK	Central Hospital Kibali
CIL	Carbon in Leach
CIM          Canadian Institute of Mining, Metallurgy and Petroleum (CIM)
CIP	Carbon in Pulp
CN	Cyanide
COS	Coarse Ore Sockpile
CP	Competent Person
CPE	Standing Committee of Evaluation
CRM	Certified Reference Material
CSR	Community Social Relations
CSS	Closed Side Setting
CTSF	Cyanide Tailings Storage Facility
CV	Coefficent of Variation
DC	Direct Current
DD/DDH   Diamond Drillhole
DMR	South African Department of Mineral Resources
DPEM	Direction de Protection de l’Environnement Minier
DRC	Democratic Republic of the Congo
DTM	Digital Terrrain Model
DTP	DTP Company, subsidiary of Bouygues
EAP	Environmental Adjustment Plan
EDA	Estimation Data Analysis
EIA	Environmental Impact Assessment
EIS	Environmental Impact Statement

EM	Electro-Magnetic
EMP	Environmental or Emergency Management Plan
EMS	Environmental Management System
EOM	End of Month
EOY	End of Year
EPS	Datamine Enhanced Production Scheduler Software
ESIA	Environmental and Social Impact Assessment
FGO	Full Grade Ore
FOS	Fine Ore Stockpile
FR	Fresh Rock
FS	Feasibility Study
FTSF	Flotation Tailings Storage Facility
FW	Foot Wall
GA	General Arrangement
GC	Grade Control
GHG	Greenhouse Gas Emissions
GM	General Manager
GPS	Global Positioning System
GT	Grade Tonnage
HAS	High-Arsenic
HDPE	High Density Polyethylene
HEP	Hydroelectric Power
HQ	Barrel Size (63.3 mm)
HR	High Recovery
HW	Hanging Wall
HY	High Yield
ICMC	International Cyanide Management
ID	Inverse Distance
IFC	International Finance Corporation
ILR	Intensive Leach Reactor
IUCN	International Union for Conservation of Nature
JORC   Joint Ore Reserves Committee (of the Australasian Institute of Mining and Metallurgy, Australian Institute of Geoscientists and the Minerals Council of Australia).
JV	Joint Venture
KCD	Karagba Chauffeur Durba Orebody
KE	Kriging Efficiency
KMS	Kibali Mining Services

18^th September 2018 Page 34

Kibali Gold Mine NI 43-101 Technical Report

KZ	KZ Structure
LAS	Low-Arsenic
LIMS	Laboratory Information Management System
LOM	Life of Mine
LR	Low Recovery
MAS	Medium-Arsenic
MASL	Metres Above (Mean) Sea Level
MBA	Master of Business Administration
MCF	Mine Call Factor
MCP	Meta-Conglomerate Package
MG	Medium-Grade
MIMMM Member of the Institute of Materials, Minerals and Mining
MO	Marginal Ore
MOTO	Moto Goldmines Limited
MPS(P)	Mineral Processing Separating (Pumping)
MRMM	Mining Rock Mass Model
MSI	3D Mine Surveying International Limited
MSO        Minable Stope Optimiser (Datamine based software for underground stope design)
MSS	Meta-Sediments
MW	Megawatt
NAG	Net Acid Generating
NQ	Core Size (47.6 mm)
OC	Open Cast
ODBC	Open Database Connectivity
OEM	Original Equipment Supplier
OFS	Optimised Feasibility Study
OK	Ordinary Kriging
OKIMO	DRC Governmental Entity
OMC	Orway Mineral Consultants
OP	Open Pit
OPEX	Operating Costs
OREAS ORE Research & Exploration Pty Ltd CRM Manufacture
OX	Oxide
PQ	Core Size (85.0 mm)
PSA	Pressure Swing Adsorption
QA/QC	Quality Assurance/Quality Control
QG	QG Australia Ltd
QKNA/KNA         Quantative Kriging Neighbourhood Analysis
QP	Qualified Person
QQ	Quantile-Quantile

RAB	Rotary Air Blasted
RAP	Resettlement Action Plan
RC	Reverse Circulation
RED	Reducing
RES	Resource Domain
RL	Elevation (m)
RMR/MRMR         Rock Mass Rating (Mean)
ROM	Run of Mine
ROMPAD         Run of Mine Pad
RWD	Raw Water Dam
RWG	Resettlement Working Group
SAMREC        South African Code for the Reporting of Exploration Results, Mineral Resources and Mineral Reserves
SAP	Saprolite or German Company
SCADA	Supervisory Control And Data Acquisition
SCH	Schist
SG	Specific Gravity
SGS	SGS Laboratories
SLTO	Social License to Operate
SMU	Selective Mining Unit
SOKIMO         Société Miniere de Kilo-Moto SA UNISARL
SOP	Standard Operating Procedure
SOX	Sarbanes Oxley
SP	Stockpiles
SQL	Structured Query Language Database
SR	Slope of Regression
SRK	Steffen Roberts and Kirsten, Engineering Company
STD/StdDev        Standard Deviation
SWATH One-dimensional analysis graph in a specific direction of interest
TDS	Total Dissolved Solids
TR	Transitional
TRANS	Transition
TSF	Tailings Storage Facility
UC	Uniform Conditioning
UFG	Ultra-fine grind
UG	Underground
UPS	Uninterruptible Power Supply
UTM	Universal Transverse Mercator
WAD	Weak Acid Dissociated
XC	Cross-Cut

18^th September 2018 Page 35

Kibali Gold Mine NI 43-101 Technical Report

2.5 Units

cm	Centimetre
ekW	Generator Output Rating in kW
g	Grammes
Ga	Billion years
g/cm3	Grammes per Cubic Centimetre
g/t	Grammes per Metric Tonne
Ha	Hectare
Kbar	Kilobar of pressure
kg	Kilogram
km	Kilometre
km2	Square kilometre
koz	Thousand ounces
kt	Thousand metric tonnes
ktpa	Thousand metric tonnes per annum
ktpm	Thousand tonnes per month
kW	Kilo Watts
m	Metre
m²	Square meter
m3	Cubic meter
Mm3	Million Cubic Metres
Ml	Million litres
Moz	Million fine troy ounces
Mt	Million metric tonnes

Mtpa	Million tonnes per annum
MVA	Mega Volt Amps
MW	Mega Watts
oz	Fine troy ounce equalling 31.10348 grams
ppm	Parts per million
t	Metric tonne
tm-3	Density measured as metric tonnes per cubic metre
°	Degrees
‘	Minutes
%	Percentage
%w/v	Percentage Weight by Volume
µm	Microns
#	Mesh
$	United States Dollar
$ ‘000	Thousand United States Dollars
$ M	Million United States Dollars
$/oz United States Dollar per ounce
$/t United States Dollars per Metric Tonne

18^th September 2018 Page 36

Kibali Gold Mine NI 43-101 Technical Report

3 Reliance on Other Experts

This report has been prepared by Randgold. For the purpose of this report, the QPs have relied upon information provided by Randgold’s Legal Counsel regarding the validity of Exploitation Permit and the changes to the fiscal regime outlined in the DRC Mining Code (2018); this opinion has been relied upon in Section 4 (Property Description and Location), Section 24 (Other Relevant Data and Information) and in the summary of this report.

18^th September 2018 Page 37

Kibali Gold Mine NI 43-101 Technical Report

4 Property Description and Location

4.1 Project Location

The Kibali Mine is a gold mining, milling and exploration project which is located in the NE of the Democratic Republic of Congo (DRC), approximately 560 km NE of the city of Kisangani and 150 km west of the Ugandan border town of Arua, near to the international borders with Uganda and Sudan. Kinshasa. The capital city of DRC, is located approximately 1,800 km SW of the Project. The location of the Project area is shown in Figure 4-1.

Access to the Project for personnel is commonly through charter flight to a local airstrip from Entebbe, Uganda, which is approximately 470 km SE. Entebbe is serviced daily by international air carriers. Road access is available from Kampala, Uganda and is approximately 650 km. The main material access points for site development and operation include the major ports of Mombasa, Kenya (1,800 km) and Dar es Salaam, Tanzania (1,950 km).

The Project, which covers an area of approximately 1,836 km2, is centred at approximately 3.13º latitude and 29.58º longitude, in the administrative district of Haut Uélé in Province Orientale.

The Project consists of multiple deposits including, Karagba-Chauffeur-Durba (KCD), Sessenge, Pakaka, Pamao, Gorumbwa, Kibali, Mengu Hill, Mengu Village, Megi, Marakeke, Kombokolo, Sessenge, and Ndala.

4.2 Mineral Rights and Land Ownership

Kibali Goldmines has been granted ten Exploitation (Mining) Permits under the the DRC Mining Code (2002) in respect of the Project, eight of which are valid until 2029 and two of which are valid until 2030.

All Mineral Resources and Ore Reserves summarised in this report are contained within these Permits (Table 4-1 provides Permit details and Table 4-2 provides Permit perimeter coordinates). The Permits occur within two territories, namely Watsa and Faradje which fall under the administrative district of Haut Uélé.

The principal mineral deposit, KCD, forms both an open pit and underground mine. This operation and the associated infrastructure (processing plant, accommodation, and airport) are within Exploitation Permits 11447 and 11467. All Kibali Exploitation Permits are presented in Table 4-1.

18^th September 2018 Page 38

18th September 2018 Page 39

Kibali Gold Mine NI 43-101 Technical Report

Table 4-1 Kibali Exploitation Permit Details

Arête No.	Permit No.	Surface Area     (km2)	Expiry Year
0852/CAB.MIN/MINES/01/2009	11447	226.8	2029
0855/CAB.MIN/MINES/01/2009	11467	248.9	2029
0854/CAB.MIN/MINES/01/2009	11468	45.9	2030
0853/CAB.MIN/MINES/01/2009	11469	91.8	2029
0329/CAB.MIN/MINES/01/2009	11470	30.6	2029
0852/CAB.MIN/MINES/01/2009	11471	113.0	2029
0331/CAB.MIN/MINES/01/2009	11472	85.0	2029
0856/CAB.MIN/MINES/01/2009	5052	302.4	2029
0858/CAB.MIN/MINES/01/2009	5073	399.3	2029
0103/CAB.MIN/MINES/01/2011	5088	292.2	2030

In the QP’s opinion, all appropriate Permits that have been acquired and obtained to conduct the work proposed for the property.

The next renewal date for the Permits are 5th November 2029 and 6th March 2030 and the current LOM plan for the Kibali Ore Reserves extends beyond these dates.

The DRC Mining Code (2002) includes provision for renewal of all Exploitation Permits for a successive period of 15 years, providing the holder has not breached the Permit obligations of Permit fee and annual surface rights fees payment and upholds all environmental standards set out in the Exploitation Permit. Furthermore, the Permit holder should provide the appropriate government departments with a monthly mining activity report and quarterly exploration reports.

The DRC Mining Code (2002) and Regulations have been amended with an updated Mining Code which came into force on 9th March 2018 (DRC Mining Code(2018)) and the related amended Mining Regulations which came into force on 8th June 2018.

All the Permit fees, surface rights fees and taxes relating to Kibali’s exploitation rights have been paid to date and the concession is in good standing.

The QPs are not aware of any risks that could result in the loss of ownership of the deposits or loss of the Permits, in part or in whole.

18^th September 2018 Page 40

Kibali Gold Mine NI 43-101 Technical Report

Table 4-2 outlines the perimeter coordinates of the Kibali Permits. All UTM coordinates use UTM Zone 35N datum WGS84 grid.

Table 4-2 Kibali Exploitation Permit Coordinates

Permit	Lat	Long	Easting	Northing	Permit	Lat	Long	Easting	Northing
5088	3°00’30”	29°51’00”	816830	332928	11472	2°57’00”	29°57’00”	827974	326501
	3°01’00”	29°51’00”	816828	333850		2°57’00”	29°56’30”	827047	326499
	3°01’00”	29°51’30”	817755	333853		2°58’00”	29°56’30”	827042	328344
	3°04’00”	29°51’30”	817741	339386		2°58’00”	29°56’00”	826115	328341
	3°04’00”	29°51’00”	816813	339384		2°58’30”	29°56’00”	826112	329263
	3°10’00”	29°51’00”	816783	350451		2°58’30”	29°54’00”	822403	329254
	3°10’00”	29°57’30”	828835	350485		2°59’00”	29°54’00”	822401	330176
	3°09’30”	29°57’30”	828838	349562		2°59’00”	29°53’30”	821474	330173
	3°09’30”	29°59’30”	832547	349573		3°00’00”	29°53’30”	821469	332018
	3°10’00”	29°59’30”	832544	350495		3°00’00”	30°01’00”	168332	332045
	3°10’00”	30°00’00”	166529	350498		2°55’00”	30°01’00”	168307	322822
	3°00’00”	30°00’00”	166477	332051	11471	3°10’00”	29°31’30”	780634	350357
	3°00’00”	29°53’30”	821469	332018		3°10’00”	29°28’00”	774147	350342
	3°00’30”	29°53’30”	821466	332940		3°19’30”	29°28’00”	774104	367859
	3°10’00”	29°31’30”	780634	350357		3°19’30”	29°31’30”	780590	367875
5052	3°19’30”	29°31’30”	780590	367875	11470	3°00’00”	30°00’00”	166477	332051
	3°19’30”	29°32’00”	781517	367878		3°09’00”	30°00’00”	166524	348653
	3°18’30”	29°32’00”	781522	366034		3°09’00”	30°01’00”	168378	348648
	3°18’30”	29°32’30”	782448	366036		3°00’00”	30°01’00”	168332	332045
	3°17’30”	29°32’30”	782453	364192	11469	2°55’30”	29°37’00”	790894	323642
	3°17’30”	29°33’00”	783380	364194		2°55’30”	29°31’00”	779770	323617
	3°16’00”	29°33’00”	783387	361428		3°00’00”	29°31’00”	779751	331915
	3°16’00”	29°36’00”	788947	361443		3°00’00”	29°37’00”	790874	331941
	3°16’30”	29°36’00”	788945	362365	11468	2°55’30”	29°31’00”	779770	323617
	3°16’30”	29°36’30”	789872	362367		2°55’30”	29°28’00”	774208	323605
	3°17’00”	29°36’30”	789869	363289		3°00’00”	29°28’00”	774189	331902
	3°17’00”	29°37’30”	791723	363294		3°00’00”	29°31’00”	779751	331915
	3°16’30”	29°37’30”	791725	362372	11467	3°10’00”	29°35’00”	787122	350373
	3°16’30”	29°38’30”	793579	362377		3°06’30”	29°35’00”	787138	343919
	3°17’00”	29°38’30”	793576	363299		3°06’30”	29°35’30”	788065	343921
	3°17’00”	29°39’00”	794503	363301		3°00’00”	29°35’30”	788093	331934
	3°18’00”	29°39’00”	794498	365145		3°00’00”	29°28’00”	774189	331902
	3°18’00”	29°40’00”	796352	365150		3°10’00”	29°28’00”	774147	350342
	3°19’30”	29°40’00”	796344	367917	11447	3°00’00”	29°37’00”	790874	331941
	3°19’30”	29°42’30”	800978	367929		3°00’00”	29°35’30”	788093	331934
	3°18’30”	29°42’30”	800983	366085		3°06’30”	29°35’30”	788065	343921
	3°18’30”	29°44’30”	804690	366095		3°06’30”	29°35’00”	787138	343919
	3°19’00”	29°44’30”	804688	367017		3°10’00”	29°35’00”	787122	350373
	3°19’00”	29°45’30”	806541	367023		3°10’00”	29°40’00”	796390	350397
	3°19’30”	29°45’30”	806539	367945		3°15’30”	29°40’00”	796364	360540
	3°19’30”	29°47’00”	809319	367953		3°15’30”	29°44’00”	803778	360560
	3°19’00”	29°47’00”	809322	367030		3°15’00”	29°44’00”	803781	359638
	3°19’00”	29°48’30”	812102	367038		3°15’00”	29°46’00”	807488	359648
	3°18’00”	29°48’30”	812108	365194		3°14’30”	29°46’00”	807491	358726
	3°18’00”	29°48’00”	811181	365191		3°14’30”	29°45’30”	806564	358723
	3°17’00”	29°48’00”	811186	363347		3°14’00”	29°45’30”	806567	357801
	3°17’00”	29°47’30”	810259	363344		3°14’00”	29°44’00”	803786	357793
	3°16’00”	29°47’30”	810264	361500		3°10’30”	29°44’00”	803803	351338
	3°16’00”	29°47’00”	809337	361497		3°10’30”	29°44’30”	804730	351341
	3°15’00”	29°47’00”	809342	359653		3°10’00”	29°44’30”	804733	350418
	3°15’00”	29°44’30”	804708	359640		3°10’00”	29°45’00”	805660	350421
	3°15’30”	29°44’30”	804705	360562		3°09’30”	29°45’00”	805662	349499
	3°15’30”	29°40’00”	796364	360540		3°09’30”	29°46’00”	807516	349504
	3°10’00”	29°40’00”	796390	350397		3°08’00”	29°46’00”	807523	346737
11472	2°55’00”	29°58’30”	830766	322819		3°08’00”	29°47’00”	809377	346742
	2°55’30”	29°58’30”	830764	323742		3°04’00”	29°47’00”	809397	339364
	2°55’30”	29°57’30”	828909	323737		3°04’00”	29°46’00”	807543	339360
	2°56’00”	29°57’30”	828906	324659		3°07’30”	29°46’00”	807526	345815
	2°56’00”	29°57’00”	827979	324657		3°07’30”	29°37’00”	790841	345772

18^th September 2018 Page 41

Kibali Gold Mine NI 43-101 Technical Report

4.3 Surface Rights

Surface rights in the area of the Kibali Permits belong to the DRC Government. Utilisation of the surface rights is granted by the Kibali Exploitation Permit under condition that the current users are properly compensated. All the license fees, surface rights fees and taxes relating to Kibali’s exploitation rights have been paid to date and the concession is in good standing.

One exclusion zone with an area of 10.26 km2 exists within the Permit surrounding the Kibali South deposit which was transferred to SOKIMO from Kibali Goldmines in December 2012. The coordinates of this exclusion zone are presented in Table 4-3.

Table 4-3 Kibali Exploitation Permit Coordinates

ID	Lat	Long	Easting	Northing
A	03°05’00”	29°35’30”	788071	341155
B	03°05’00”	29°34’00”	785291	341148
C	03°06’00”	29°34’00”	785286	342992
D	03°06’00”	29°32’30”	782506	342986
E	03°05’30”	29°32’30”	782508	342064
F	03°05’30”	29°33’30”	784361	342068
G	03°04’00”	29°33’30”	784368	339302
H	03°04’00”	29°35’30”	788076	339311

The QPs are not aware of any other significant factors and risks that may affect access, title, or the right of ability to perform work on the property.

4.4 Ownership, Royalties and Lease Obligations

Kibali Goldmines is owned 90% by a joint venture between Randgold and AngloGold Ashanti (45/45), and 10% by Congolese parastatal Société des Mines d’Or de Kilo-Moto (SOKIMO). SOKIMO is wholly owned by the DRC with the shareholding held by the Minister of Portfolio of DRC (MoP). SOKIMO is wholly owned by the DRC with the shareholding held by MoP. OKIMO was transformed into SOKIMO in December 2010.

Randgold is the operator at Kibali for both exploration and mining.

The DRC Mining Code (2002) and Regulations have been amended with an updated Mining Code which came into force on 9th March 2018 (DRC Mining Code(2018)) and the related amended Mining Regulations which came into force on 8th June 2018.

The following changes have been made to the DRC Mining Code (2002) that could have an impact on Kibali:

☐ Royalty charges are to be increased from 2.5% to 3.5%. This increases royalty charges over the LOM by an estimated $94.5 M, which would not materially impact the LOM profitability.

☐ Various increases in import and other duties from 4% to 7% depending on consumable type, which would not materially impact the LOM profitability.

18^th September 2018 Page 42

Kibali Gold Mine NI 43-101 Technical Report

☐ A super-tax profit has been promulgated based on the feasibility study prepared at the time the approval was given for the building of the Project. Accordingly, such a tax would only apply if the average annual gold price was in excess of $2,000/oz.

The exact impact, if any, of the changes will only be fully known once the 2018 Mining Code and related regulations are clarified and implemented in full.

Going forward the DRC Mining Code (2018) envisages a stability period for the tax, customs and exchange control regime of five years from the date on which the DRC Mining Code (2018) came into force and further provides that a number of the taxes shall be applied in accordance with the applicable substantive law.

Kibali Jersey Limited, the holding company of Kibali, the shareholders of Kibali Jersey Limited and Kibali Gold Mines SA, are considering all options to protect their vested rights under the DRC Mining Code and to enforce the additional state guarantees previously received, including preparations for international arbitration. In addition, engagement with the DRC government is ongoing, with the aim of exploring alternative solutions, which could be mutually acceptable to both parties. This includes the application of Article 220 of the DRC Mining Code (2018), which affords benefits to mining companies such as Kibali, operating in landlocked infrastructurally challenged provinces. If Article 220 were applied to Kibali, any advantages granted would mitigate any impact of the implementation of the DRC Mining Code (2018).

The QP notes that the mining code is in the process of transition and the current proposed changes do not have an impact on the stated Mineral Resource or Ore Reserves at the gold prices used for Ore Reserve ($1,000/oz), Mineral Resource ($1,500/oz), or the current gold prices of $1,200/oz (August 2018).

The QP is not aware of any risks that could result in the loss of ownership of the deposits or loss of the Permits, in part or in whole.

18^th September 2018 Page 43

18th September 2018 Page 44

Kibali Gold Mine NI 43-101 Technical Report

5 Accessibility, Climate, Local Resources, Infrastructure and Physiography

5.1 Accessibility

Kibali is located in the NE of the DRC, approximately 560 km NE of the city of Kisangani and near the international borders with Uganda and South Sudan.

The Project area is situated in a rural setting that lacks local infrastructure. Infrastructure in the DRC is generally poor as a result of limited investment in the maintenance of road networks established during colonial times. Historically, the lack of investment was the result of civil unrest and diminished government revenue collection.

The main access points for equipment and supplies for the operation include the major ports of Mombasa, Kenya (1,800 km) and Dar es Salaam, Tanzania (1,950 km). The routes are paved up to the DRC border. Road access is from Kampala, Uganda and is approximately 650 km. The arterial road between Arua and site is unpaved but has been upgraded and serves as the main access route for materials to site. Local roads are generally in very poor states of repair. Supplies typically require two weeks to arrive from Mombasa.

A local certified airstrip with passport control, serves as the primary access point to site for personnel on charter flights from Entebbe, Uganda, which is approximately 470 km SE of the Mine. Entebbe is serviced daily by international air carriers.

5.2 Climate

The DRC has a total area of 2.3 million km². The country straddles the equator and is characterised by dense tropical rain forest in the central Congo River basin and highlands in the east.

The climate is tropical - hot and humid in the equatorial river basin; cooler and drier in the southern highlands; cooler and wetter in the eastern highlands where Kibali is located.

The Watsa territory wet season occurs between March to November, with the dry season occurring between November and March (Figure 5-1). Watsa experiences extreme seasonal variation in monthly rainfall with most rain occurring in heavy tropical thunderstorms. Precipitation is highest in October, and January and December are the driest months. Humidity levels are highest in the wet season.

18^th September 2018 Page 45

Kibali Gold Mine NI 43-101 Technical Report

Figure 5-1 Kibali 10 Year Rainfall Statistics by Monthly Actual Measurement

The Watsa territory dry season lasts from January to March, with average daily high temperatures above 30°C and average daily low temperatures approximately 19°C. The cool season occurs between May and November, with average daily high temperatures below 29°C and average daily low temperatures approximately 18°C.

The average wind speed experiences mild seasonal variation over the course of the year, generally averaging 8.0 km/h in the wet season and 6.5 km/h in the dry season.

Climatic conditions do not materially affect either exploration, development, or mining operations.

5.3 Local Resources

The mine office of Kibali Goldmines is located in the village of Doko, which is centrally located within the Project area and approximately 180 km by road from Arua on the Ugandan border. The district capital of Watsa lies about 9 km to the south of the Project, which is situated just north of the Kibali River on the road to Faradje and the Sudan. The town of Bunia, which is the United Nations controlled entry point to NE DRC, lies about 200 km to the south of the Mine.

The population in the Kibali area is approximately 65,000. The Watsa territory population is approximately 300,000. Figure 5-2 presents a plan view of the Kibali deposits and surrounding communities.

18^th September 2018 Page 46

Kibali Gold Mine NI 43-101 Technical Report

There are generally limited services in the area that are suitable to directly support the Kibali mine as such Kibali has completed extensive infrastructure upgrades during construction.

As per Randgold’s strategy, in 2017, the Mine continued to focus on host country employment and skills transfer, steadily increasing the Congolese component to approximately 92% of the full time Kibali manpower. Congolese contractors are also utilised for construction projects and for ore haulage. The Mine has assisted, and continues to assist, local start-up businesses.

5.4 Infrastructure

The local Project area lacks any substantial infrastructure to support the mining operation, other than that which has been constructed by Kibali. All existing infrastructure supports the local subsistence and small-scale agriculture.

Remnants of historical mining activities can be found on the property (residential buildings, processing plant, underground mine shafts, and surface workings) in various states of repair. Although remnants of the historical mining activities remain, the mine is essentially a greenfield development, with new facilities having been built to support the current mining and processing activities, because the current mine is of a much larger scale than any of the historical mining infrastructure.

The key on-site surface and underground infrastructure at Kibali include the following:

☐ Mine access and internal road network.

☐ A 7.2 Mtpa process plant.

☐ Two TSFs; one lined facility for the Oxide (CIL) tails and one unlined facility for the Sulphide (Flotation) tails.

☐ Accommodation village for married and single staff and employees.

☐ Administrative buildings, stores warehouses, laboratory, workshops for surface and underground equipment, security buildings, medical and emergency response facilities.

☐ Fuel Storage.

☐ Raw and process water containment and storage dams and water distribution network.

☐ Communications and data transmission networks.

☐ Airstrip.

☐ Twin declines and vertical production shaft and a series of ramp-connected levels.

☐ Diesel generator station installed with CAT 3516B-HD (1.5 MW) generators.

18^th September 2018 Page 47

18th September 2018 Page 48

Kibali Gold Mine NI 43-101 Technical Report

There is no national grid power supply to the site, as a result Kibali is totally dependent on its own generation facilities.

The power supply currently comes from a mix of on-site, high-speed diesel generator sets and two off-site hydropower stations; Nzoro II is currently producing 22 MW and Ambarau produces 10MW. A further hydropower station is under construction at Azambi. When Azambi has been commissioned, total peak hydropower capacity will be 42 MW, which is sufficient to meet the mine power demand. The site is connected to the hydrostations via a 66 kV overhead line network.

The diesel generators currently are supplying approximately 7 MW with hydro providing the majority of the demand.

The primary source of raw water supply is rain and spring water catchments with top-up from a borehole system and a final backup from the Kibali River. Raw water is collected and stored in the raw water dam, which has a storage capacity of 9,500 m3. The processing plant requires approximately 46,000 m3 of water per day, which is sourced by reclaiming water from the Flotation Tailings Storage Facility (FTSF) and CTSF1 and CTSF2.

Figure 5-3 presents a panoramic overview photograph of Kibali (looking west).

5.5 Physiography

The topography of the area is gently hilly, ranging in elevation between 700 m to 1,500 m above sea level (MASL). The immediate Project area is characterised as generally hilly, which includes several discrete hills up to 170 m high. The plant site is located on a flat plain area which lies at about 860 MASL. The Project lies in a low seismic rated area.

Vegetation is dominated by elephant grass with forested areas along drainages. It is likely that the entire area comprised rainforest prior to modification by human activity.

18^th September 2018 Page 49

18th September 2018 Page 50

Kibali Gold Mine NI 43-101 Technical Report

6 History

6.1 Historical Exploration and Development

The discovery of gold in the region is attributed to Hannan and O’Brien in 1903. Historical gold production from the Kilo and Moto areas between 1906 and 2009 is estimated to be approximately 11 Moz half of which came from alluvial deposits. Mining operations were conducted by the Belgian Government via the Société des Mines d’Or de Kilo-Moto (SOKIMO), which was established in 1926. Most of the mining activity within the Project area was undertaken during the 1950s but accurate production records have been lost over the years of civil unrest in the region. Gorumbwa, Agbarabo and Durba deposits are believed to have produced more than 60% of the over 3 Moz of recorded gold production from the Moto area. The SOKIMO processing plant was located near the old Durba mine. The plant comprised crushing and ball milling circuits, followed by gravity, cyanide leach and mercury amalgamation circuits.

After independence in 1960, gold production dropped sharply as mining was mainly undertaken by artisanal workers and small-scale alluvial operations. SOKIMO changed its name to Offices des Mines d’Or de Kilo-Moto (OKIMO) in 1966 and was the main operator in the Project area. Sporadic underground mining was conducted in the Project area after 1960, however this is believed to be of a remnant nature and as such negligible amounts of gold were produced. Accurate production records are not available due to the civil unrest in the region during the 1980s and 1990s.

Davy McKee undertook a detailed assessment of the area on behalf of the Government of Zaire in 1991, with funding from the African Development Bank. This assessment included a significant amount of drilling to verify historical data.

Barrick Gold Corporation (Barrick) acquired exploration rights over most of the Kilo-Moto belts in 1996 in a 70/30 joint venture with the government entity OKIMO and drilled a number of targets as well as completing regional and detailed soil sampling programs. Subsequently Barrick formed a joint venture (JV) with AngloGold Ashanti to split equally their 70% holding of the Project.

Kibali was discovered by the Barrick and AngloGold Ashanti JV in 1998 and AngloGold Ashanti became the operator of the Project. The Barrick and AngloGold Ashanti JV completed a number of drilling programs, mainly concentrated at Kibali and Pakaka. The Barrick and AngloGold Ashanti JV also carried out soil sampling over most of the concession area, and a regional aeromagnetic survey was completed by World Geoscience Limited (WGC). The survey was undertaken at 200 m line spacings and the data was interpreted by WGC. AngloGold Ashanti and Barrick withdrew from the Project in 1998 due to local unrest and civil war. Very little information is available regarding drilling by OKIMO or the Barrick and AngloGold JV.

Moto Goldmines Limited (Moto) acquired the available 70% stake in the Project in 2004. Moto completed a pre-feasibility study in 2006, a Feasibility Study in December 2007, and an Optimised Feasibility Study in March 2009.

18^th September 2018 Page 51

Kibali Gold Mine NI 43-101 Technical Report

In July 2009, Randgold and AngloGold Ashanti entered into a 50/50 JV, which acquired Moto and their 70% ownership of the Project. In December 2009, the JV acquired an additional 20% shareholding of Kibali from SOKIMO. The DRC State remained a partner in the Project through OKIMO retaining a 10% interest.

Table 6-1 presents a summary of the known historical trenches, auger, and pit exploration results at Kibali.

Table 6-1 Summary of Historical Kibali Trenches, Auger and Pits Summary

Year	Company	Trenches		Auger		Pits		Total
		Meters	No.	Meters	No.	Meters	No.	Meters	No.
1950 to 1960	OKIMO	167	9	-	-	1,144	79	1,311	88
1980	MOTO	No Information Available
1996	Barrick – AngloGold Ashanti	No Information Available
2006 to 2007	MOTO	-	-	-	-	12	2	12	2
2008 to 2009	MOTO	-	-	260	135	-	-	260	135
	Total	167	9	260	135	1,156	81	1,583	225

6.2 Kibali Project Milestones and Development

The key milestones of the Kibali Goldmines’ Project are described below:

2006

☐ On site due diligence by Randgold.

2009

☐ Randgold and AngloGold Ashanti entered into a 50/50 JV in July under which they agreed to make an offer for the entire issued share capital of Moto Goldmines (70% Kibali).

☐ The JV acquired a further 20% in Kibali from SOKIMO in December, resulting in Randgold and AngloGold Ashanti each having a 45% interest in Kibali Goldmines SA respectively. The DRC State remained a partner in the Project through OKIMO retaining a 10% interest.

☐ Kibali Goldmines initiated exploration activities.

2010

☐ Feasibility study updated, and Ore Reserves doubled to over 10 Moz of gold.

☐ Formal notice to illegal miners and work programme started in March.

☐ Memorandum of Understanding (MOU) signed with Catholic Church in July.

2011

☐ Established Kokiza village for resettlement in February.

☐ Revised Feasibility Study issued in August.

☐ Preconstruction started in October.

☐ Optimised Feasibility Study issued in November.

18^th September 2018 Page 52

Kibali Gold Mine NI 43-101 Technical Report

2012

☐ JV approves Feasibility Study and approval of capital for development in May.

☐ KCD open pit mining begins.

☐ Commissioning and ramp up of metallurgical facility.

☐ Twin decline development headings started.

☐ Shaft sinking initiated.

2013

☐ Associated infrastructure completed.

☐ First gold produced and completion of Resettlement Action Plan (RAP) in September.

☐ Oxide circuit in operation.

2014

☐ Nzoro 2 hydropower station completed.

☐ First production from underground stopes.

2015

☐ Full year of open pit production at design specifications.

☐ Completed shaft sinking and commenced equipping.

☐ Backfill plant commissioned.

☐ Initiated production from Mengu Hill.

2016

☐ Underground ramp up.

☐ Three new satellite open pits begin production (Kombokolo, Pakaka, and Rhino).

☐ Mill throughput ramp-up.

☐ Commenced construction of Azambi hydropower station.

☐ Completed construction of Ambarau hydropower station.

2017

☐ First power from Ambarau hydropower station in the 1st Quarter.

☐ Shaft and underground system commissioned with hoisting ramp-up in the 4th quarter.

☐ Completed ultra-fine-grind and associated circuit expansion.

6.3 Historical Resource and Reserve Estimates

The following estimates are considered to be historical in nature and should not be relied upon. A Qualified Person has not completed sufficient work to classify the historical estimate as a current Mineral Resource or Ore Reserve and Randgold is not treating the historical estimates

18^th September 2018 Page 53

Kibali Gold Mine NI 43-101 Technical Report

as current Mineral Resources or Ore Reserves. They have been superseded by the Mineral Resource and Ore Reserve estimates in this report.

Table 6-2 and Table 6-3, respectively, list the historical Moto Mineral Resource and Ore Reserve estimates as of August 2008. These historical estimated Mineral Resources and Ore Reserves were classified and reported in accordance with the 2004 Australasian Code for Reporting of Mineral Resources and Ore Reserves (2004 JORC Code).

Table 6-2 presents a tabulation of the Mineral Resources within the Moto Gold Project, estimated above a nominal 1.0 g/t Au cut-off within the interpreted mineralised domains for the Moto Gold Project as of August 2008.

Table 6-2 Moto Goldmines Ltd. Mineral Resource Estimate as of August 2008

	Indicated Mineral Resources			Inferred Mineral Resources
Deposit	Tonnes (Mt)	Gold Grade (g/t Au)	Contained Gold (Moz)	Tonnes (Mt)	Gold Grade (g/t Au)	Contained Gold (Moz)
Pakaka	16.9	2.5	1.4	-	-	-
Gorumbwa	-	-	-	8.3	5.2	1.4
Kibali	-	-	-	17	2.2	1.2
Mengu Hill	8.8	3.0	0.8	-	-	-
Mengu Village	1.2	1.9	0.07	0.08	1.4	0.004
KCD	67	3.6	7.7	74	3.4	8.1
Megi	-	-	-	4.1	2.1	0.3
Marakeke	-	-	-	2.4	1.7	0.1
Kombokolo	2.3	2.4	0.2	-	-	-
Sessenge	8.6	2.3	0.6	-	-	-
Ndala	-	-	-	0.3	4.0	0.03
Pamao	7.9	1.9	0.5	1.2	1.9	0.07
Total	112	3.1	11	107	3.3	11

Table 6-3 presents a tabulation of the Probable Ore Reserves estimated within the Moto Gold Project, based upon an Optimised Feasibility Study (OFS) pit designs, is 31 Mt at 3.2 g/t for 3.2 Moz of gold. These Ore Reserves are contained within the Indicated Mineral Resources of the Moto Gold Project.

Table 6-3 Moto Goldmines Ltd. Ore Reserve Estimate as of August 2008

Pit	Tonnes  (Mt)	Gold Grade  (g/t Au)	Contained Gold (Moz)
KCD	14	3.6	1.6
Kombokolo	0.5	3.0	0.05
Mengu Hill	5.4	3.4	0.6
Pakaka	6.1	2.7	0.5
Pamao	1.5	2.1	0.1
Sessenge	2.9	2.5	0.2
Total	31	3.2	3.2

18^th September 2018 Page 54

Kibali Gold Mine NI 43-101 Technical Report

Figure 6-1 presents a chart of Kibali’s Mineral Resource and Ore Reserve evolution from Randgold’s initial acquisition to year end 2017.

Figure 6-1 Kibali Mineral Resource and Ore Reserve Evolution

6.4 Past Production

Since commencing mining operations in 2012 to the end of 2017, a total of 161 Mt (30 Mt ore), have been mined from the various deposits at Kibali. Table 6-4 summarises the past mill production for the Kibali mine.

Table 6-4 Past Production Records for the Kibali Mine

Year	Tonnes Milled (kt)	Grade (g/t Au)	Contained Gold (oz)	Recovery (%)
2013	808	3.7	88,200	91.3
2014	5,568	3.7	526,627	79.3
2015	6,833	3.5	642,720	83.8
2016	7,296	3.1	585,946	80.0
2017	7,619	2.9	596,225	83.4
Total	28,124	3.3	2,439,718	82.3

The historical gold production by previous operators and artisanal miners is unknown.

18^th September 2018 Page 55

Kibali Gold Mine NI 43-101 Technical Report

7 Geological Setting and Mineralisation

7.1 Regional Geology

The Kibali deposits are hosted within the Kibali Greenstone Belt, bounded to the north by the West Nile Gneiss and to the south by plutonic rocks of the Watsa district (Figure 7-1). The belt comprises three lithostratigraphically distinct blocks. Psammo pelitic schists, amphibolite, banded iron formations, and gneissic granitoid sills metamorphosed under upper greenschist to low-mid-amphibolite facies conditions form the eastern part of the belt. Relatively weakly foliated basalts, cherts, siliciclastic rocks, dacitic volcaniclastic rocks, and carbonaceous argillite metamorphosed under mid to upper greenschist facies conditions comprise the central and western-most parts of the belt. Granitoid plutons as old as ca. 2640 Ma intrude these rocks. A thick package of immature sandstone, gritstone, conglomerate, and probably acid tuffs forms much of the western part of the belt, including the host rocks to KCD, the largest deposit discovered to date within the belt. Radiometric dating indicates these siliclastic rocks were deposited during a belt-wide basin extension event between ca. 2629-2626 Ma, with much of the detritus derived from adjacent older parts of the belt.

The Kibali Greenstone Belt is an elongate WNW-ESE trending terrane containing Archean aged volcano-sedimentary conglomerate, carbonaceous shales, siltstone, banded iron formations, sub aerial basalts, mafic intermediate intrusions (dykes and sills) and multiple intrusive phases that range from granodiorite, tonalite and gabbroic in composition. Based on textures and types of lithologies present in the stratigraphy the rocks within the Project area are interpreted as being laid down in an aqueous environment.

The belt is bounded to the north by the West Nile Gneiss complex, a Meso-Archean granite gneiss that extends northward into the Sahara Desert. To the south the belt is bounded by the Upper Zaire Granitic Massif, an Archean granite-gneiss terrane that dominates the NE Congo Craton. The Massif is locally represented by the Watsa Igneous Complex.

The majority of the primary lithologies are clastic (sedimentary) in origin, possibly being developed in a regional extensional environment such as a rift graben or half graben. The Kibali deposits are predominantly hosted within sedimentary lithologies that have undergone complex structural deformation and metamorphism. Metamorphic grade varies from lower greenschist facies in the west, progressively increasing to amphibolite facies in the east.

Intrusive units from both the West Nile Gneiss and Kibali Greenstones are bimodal in geochemistry, with negative trending trace element distribution indicating formation in an island arc environment. The similarity of trace element distribution interpreted as representing a common origin for the intrusive units. Extrusive units from both terranes show a flat trace element signature that is more typical of MORB (Mid Oceanic Ridge Basalts). “Pillow” textured basalts are noted in the Project area.

18^th September 2018 Page 56

18th September 2018 Page 57

Kibali Gold Mine NI 43-101 Technical Report

Regional geologic interpretations suggest that the belt is a thrust stack that developed during the collision of an island arc along the northern margin of the Upper Zaire Granitic Massif with the West Nile Gneiss thrust southward over the Kibali greenstone belt. Ductile and brittle deformation events are observed in the lithological units, with polyphase isoclinal and recumbent folding mapped in some of the deposits. The belt is cut by two principal structure sets: NW -SE striking, NE dipping thrust faults and a series of sub-vertical NE-SW shear structures both of which in association with the folding are considered important mineralising controls.

7.2 Structural Geology

Gold deposits of the Kibali district are scattered along a curvilinear zone ca. 20 km long and up to a km wide known as the KZ Structure (KZS). Gold is concentrated in gently NE to NNE-plunging shoots whose orientations are generally parallel with a prominent lineation in the mineralised rocks. It has been concluded that the structure of the Kibali district is the product of at least seven phases of deformation. Key features of each event are listed below:

☐ D_1/14: ductile faults generally parallel to lithologic layering but which locally cut across lithologic layering.

☐ D_2/14: isoclinal recumbent folds whose axial planes dip ca. 25° to 30° NNE, axes that plunge ca. 25° NE, and an associated generally layer-parallel foliation.

☐ D_3/14: upright folds whose axial planes dip steeply towards either the NW or SE and axes plunge ca. 25° NE.

☐ D_4/14: sericite-rich spaced foliation largely restricted to altered rocks at KCD.

☐ D_5/14: NE-striking steeply dipping brittle faults close to parallel with axial planes of the F3/14 folds.

☐ D_6/14: SSW-dipping folds with near horizontal axes that trend WNW or ESE, an associated axial plane-parallel crenulation cleavage, and related contractional faults.

☐ D_7/14: Minor SSW-dipping normal faults, fractures, and associated barren en-echelon quartz veins.

Mineralised lodes formed at some time between the S_4/14 sericite foliation, which they overprint, and movement on the D_5/14 faults. The D_2/14, D_5/14, and D_6/14 events described here correlate with the D1, D2, and D3 events of Davis (2004) respectively.

The prominent gently NE-plunging lineation widely developed throughout the KZ Structure and whose origin has been the subject of much discussion is parallel to the axes of both the F_2/14 and F_3/14 folds. The lineation marks the intersection of F_2/14 and F_3/14 axial planes with earlier fabrics such as the S_2/14 and S_1/14 foliations and lithologic layering. It is therefore not parallel with the tectonic transport direction during either phase of deformation. NE-plunging tube-shaped folds previously referred to as sheath folds examined during this program are refolded coaxial folds and not therefore indicative of an episode of exceptionally high strain along the KZS.

D_1/14 through D_4/14 are all ductile in character. Each involved the formation of ductile faults, folds, penetrative foliations, and/or penetrative linear fabrics. D_2/14 and D_3/14 clearly occurred in a contractional setting, but evidence of the tectonic setting of D_1/14 and D_4/14 are more ambiguous. D_5/14 is a phase of essentially brittle faulting that was followed by a return to a more ductile style

18^th September 2018 Page 58

Kibali Gold Mine NI 43-101 Technical Report

of contractional deformation during D_6/14. The D_7/14 event likely represents some type of minor tectonic relaxation following cessation of D_6/14 shortening.

Most aspects of the district-scale structural architecture formed during D_1/14 -D_3/14 although the effects of D_5/14 faulting are locally apparent in district-scale geological maps. The contact between the West Nile Gneiss and Archean rocks north of Kibali is approximately parallel to the strike of F_6/14 fold axial planes and D_6/14 reverse faults. The limited vergence data recorded for F_6/14 folds indicates top to the SSW displacement, the same direction in which the West Nile Gneiss is Inferred to have been thrust onto the Archean rocks of the Kibali district, suggesting D_6/14 is that event. This may also be the early Palaeozoic event that disturbed radiogenic Pb-isotopes in zircon and monazite throughout the Kibali district.

7.3 Local Geology

The Kibali deposits differ from many orogenic gold deposits in terms of structural setting. Rather than being linked to a major large scale steeply dipping strike slip fault with brittle-ductile deformational evolution, they are hosted within a thrust stack sequence with ductile to brittle-ductile deformational structures and complex folding history. Two main structure sets characterise the Project area, NW-SE trending NE dipping thrust faults which have produced recumbent folds and some repetition of the stratigraphy; and a series of sub-vertical NE-SW trending shear and fold structures locally termed S2 structures which contributed to the formation of and deformation of early folds to create localised zones of refolded fold or sheath folding. The S2 structures may be older fold or basinal structures exploited by reactivation. The Project area is cut by regional scale NE trending faults that are both pre and post mineralisation (Figure 7-2). A structural analysis of drill core and outcrop in the Project area has identified an early but regionally consistent foliation (S1), defined by the preferred alignment of phyllosilicates and strongly shear-attenuated clasts (Figure 7-3). This early compression-related fabric is sub-parallel to bedding contacts and generally has a north-west strike and a shallow NE dip. Davis (2004) identified this fabric as a shear fabric based upon interpreted ‘asymmetric lozenges and relict fold hinges’ and an L1 defined as a ‘stretching lineation’. The stretching lineation lies within the plane of the S1 foliation and attains up to a 10:1 elongation ratio in the NE plunge direction. The stretching is interpreted by Davis (2004) as being the result of shearing (simple rotational shear) with possible direction of tectonic transport being either to the NE or to the SW. A localised, S2 foliation has an approximate NE strike and dips sub-vertically, either NW or to the SE. The S2 fabric is constrained to the NE trending transfer S2 shear corridors, and ongoing structural mapping from open pit exposures are suggesting that they may be axial plane cleavage developed in very early folds with NE plunging axial planes. The model is still under investigation. D3 fabrics are pervasively developed throughout the Project area and are observed in outcrop and drill core as a delicate crenulation cleavage (S3), which is best developed in the finer-grained sedimentary rocks

18^th September 2018 Page 59

18th September 2018 Page 60

Kibali Gold Mine NI 43-101 Technical Report

Figure 7-3 Photograph Showing Inter-Bedded Carbonaceous Argillite. Shale and Siltstone with Sub-parallel S0 and S1, Overlain by Transgressive Dissolution Crenulation Cleavage (S3)

the clastic sedimentary conglomerate is a matrix supported, polymictic and monomictic fragmental unit with sub-angular to sub-rounded clasts of banded chert, jasper, siltstone, such as the argillite and mudstone units. The D3 fabric and structures refracts into the more siliceous and coarser grained units where they become difficult to observe. Occasional banded iron formation (BIF) fragments, argillite and rare mafic volcanic set in a poorly-sorted chloritic matrix (Figure 7-4). The chlorite reflecting the regional green schist facies metamorphism, which becomes paler in colour proximal to the albite-carbonate-silica (ACSA) alteration associated with mineralisation. The unit displays both fine-grained and coarse-grained varieties. This unit may have originated as a mass flow deposit or as a tuff-breccia unit within the basin development. The unit is the typical host to the ACSA and auriferous pyrite, due in part to the competency contrast with proximal carbonaceous argillite-shale and ironstone units during deformation.

Figure 7-4 Photograph Showing the Coarse Clastic Conglomerate Unit with Sub-Parallel S0 and S1

18^th September 2018 Page 61

Kibali Gold Mine NI 43-101 Technical Report

7.4 Mineralisation

Mineralisation Style

Gold deposits of the Kibali district are part of the globally significant group of Neoarchean orogenic gold deposits, examples of which are found in most Neoarchean cratons around the world. At Kibali the gold deposits are largely hosted in siliciclastic rocks, banded iron formations, and chert that were metamorphosed under greenschist facies conditions. Ore-forming H₂O-CO₂-rich fluids migrated along a linked network of gently NE dipping shears and NE to NNE-plunging fold axes that is commonly referred to as the KZ Trend. On-going deformation during hydrothermal activity resulted in development of lodes in a variety of related structural settings within the KZ Trend. The source(s) of metal and fluids which formed the deposits remain unknown, but metamorphic devolatilisation reactions within the supracrustal rocks of the Moto Greenstone Belt and/or deeper fluid and metal sources may have contributed.

The preliminary mineralisation model for the area suggests ore-forming fluids were produced in a convergent tectonic environment as part of a thickening thrust stack. Progressive metamorphism and devolitisation of the lower stack generated fluids which ascended upwards along faults, scavenging sulphur and metals. The fluids migrated upward and southward along NE dipping thrust faults and NE trending S2 shears. The S2 shears contributed to development of sheath folding, which in turn contributed to the formation of ACSA in proximal host rocks. The ACSA shattered with progressive deformation, allowing further infiltration of fluids and deposition of gold and sulphides (pyrite). The alteration varies in intensity from weak to texturally destructive (note: petrographic studies indicate that the level of albite in the ACSA assemblage is variable, the main constituents being ferroan carbonate (siderite/ankerite), silica, chlorite, sericite, and sulphide).

Mineralisation Characteristics

Gold deposits of the Kibali district are associated with halos of quartz, ankerite, sericite, ± albite (ACSA-A) alteration that extend for 10s to 100s of metres into the adjacent rocks. This widespread ACSA-A alteration assemblage is superimposed on older greenschist facies metamorphic assemblages. Locally, in the vicinity of the main mineralised zones, ACSA-A alteration is overprinted by ankerite-siderite, pyrite alteration (ACSA-B) that hosts the ore. Gold is directly associated with the ACSA-B alteration assemblage. In smaller peripheral deposits a late chlorite, carbonate, pyrite assemblage is associated with the ore rather than the ACSA-B assemblage, implying a district-wide zonation of mineral assemblages along and across the mineralised KZ-Trend. Zones of auriferous ACSA-B alteration are commonly developed along the margins of banded iron formations, or contacts between chert, carbonaceous phyllite, and banded iron formations. Mineralised rocks in the Kibali district typically lack significant infill quartz-rich veins, unlike many other orogenic gold deposits. Gold is instead associated with pyrite in zones of alteration that replaced the earlier mineralogy of the host rocks. Local remobilisation and upgrading of ACSA-B related ore occurred adjacent to the margins of some post-ore crosscutting chlorite, carbonate, ± pyrite, ± magnetite-altered diorite dikes.

18^th September 2018 Page 62

Kibali Gold Mine NI 43-101 Technical Report

The majority of gold mineralisation hosted by the KCD and satellite deposits is texturally associated with fine disseminated pyrite, with minor pyrrhotite and arsenopyrite. The auriferous pyrite occurs as both ‘salt and pepper’ disseminated fine grains and clusters of disseminated grains forming blebs and pseudo-vein mosaics. Petrographic study has identified several sulphide phases with arsenopyrite, chalcopyrite, pyrrhotite and pyrite dominating the assemblage with multiple generations of each identified. Gold is hosted within the dominant second pyrite phase and as late fracture fillings associated with chalcopyrite and galena. The gold bearing pyrite is hosted by the sequence of coarser clastic sedimentary unit’s conglomerate and chert-ironstone assemblage, often with an envelope of ACSA (Figure 7-5).

It should be noted that some gold mineralisation does occur in units with a lack of enveloping ACSA. Increased grades are associated with strong ACSA with disseminated sulphides – the observation has been interpreted as being a result of silicified and altered host units becoming brecciated as deformation progressed, producing competency contrasts, and increasing permeability. A similar setting is interpreted for the host lithologies, where coarser clastics and chert/ironstone units seem to have behaved in a more brittle fashion (finer grained sediments behaved more ductile).

Figure 7-5 Photograph Showing Pervasive Primary Lithology Texture Destructive Alteration ACSA Which Has Been Cut by Late-Stage Silica and Silica-Pyrite Veins

18^th September 2018 Page 63

Kibali Gold Mine NI 43-101 Technical Report

7.5 Project Deposits

The majority of the mineralisation currently being delineated at the Project occurs within two broad mineralised corridors with deposits and prospects along each corridor (Figure 7-6). The first is a NE trending structural-alteration corridor trending called the Sessenge-KCD Trend. This corridor has been interpreted as being coincident with a graben or half graben and is cut by several NE trending S2 structures. The second is a NW trending corridor that stretches from the Pakaka deposit in the SE to the Mengu Hill deposit in the NW and is called the Pakaka-Mengu Trend. The corridor is interpreted as representing the surface expression of one of the regional NW trending D1 thrust faults, with elevated mineralisation occurring at the intersection with NE trending S2 corridors.

Karagba-Chauffeur-Durba (KCD) Deposit

The KCD deposit is the principal mineralised occurrence along the Sessenge-KCD Trend and consists of three semi-vertically stacked lodes hosted within the volcano-sedimentary units with mineralisation showing a strong correlation with texture destructive quartz-ankerite-chlorite alteration (Figure 7-7). The lodes are broadly categorised as the upper 3000 lodes, 5000 lodes, and at the deeper 9000 lodes. All generally plunge from surface to the NE at low to moderate angles (approximately 30°) with mineralised wireframes based on drilling intercepts indicating a down plunge continuation of approximately 2,000 m (remaining open down plunge).

The 3000 lode crops out in the present open pit (Karagba) and is the western-most lode. It is approximately 300 m in width, 30 m thick, and has a broad gentle and open semi-synclinal form to its plunge.

18^th September 2018 Page 64

18th September 2018 Page 65

18th September 2018 Page 66

Kibali Gold Mine NI 43-101 Technical Report

The 5000 lode outcrops slightly east and south of the 3000 lode (Chauffeur and Durba) and forms the majority of the topographically elevated area known as the Durba Hill, on which the historic Durba Plant is situated. The lodes are more sub-vertical in attitude than the 3000 and 9000 lodes and are consistently of higher grade.

The 9000 lode does not outcrop in the KCD open pit but crops out to the south of the Durba Hill at Sessenge. The 9000 lode is comprised of two main lodes 9101 and 9105. The 9105 is of a similar shape and attitude as the 5000 lode, with the 9101 joining Sessenge and is a shallow dipping lens with a similar plunge to the 5000 lode.

The lodes of the KCD deposit show a strong spatial association to the NE trending S2 structures. In simplified terms the lodes may be linked genetically by large-scale recumbent folding developed between two bounding NE trending structures locally termed the Eastern Transfer Fault and the Western Transfer Fault. It should be noted that almost all of the anomalous and economic mineralisation at the KCD deposit occurs in areas located between the Eastern Transfer Fault and and Western Transfer Fault, in conjunction with increases in alteration and structural deformation and localised refolded fold (previously called sheath fold) development. Mineralisation is hosted with conglomerate units, and ironstone and chert assemblages, enveloped within a halo of weak to moderate albite-siderite-ankerite-silica-sericite-sulphide alteration. Higher grade (with increased sulphide content) developed in zones of strong to intense alteration that overprinted and texturally destroyed previous breccia, foliation, and lithological textures.

The location of the individual lodes within the KCD deposit are intimately controlled by the position, shape, and orientation of a series of gently NE -plunging tight to isoclinal folds. The ACSA-A alteration developed during the formation of these folds, and the sericite foliation which is an integral part of the ACSA-A assemblage formed parallel to their axial planes. Zones of later auriferous ACSA-B alteration developed along the axis, limbs, and more rarely the axial planes of these folds, locally wrapping around the hinges of the folds to form elongate NE plunging concave-shaped rods. ACSA-B alteration is also commonly focused along the margins of more extensive banded iron formations, indicating a stratigraphic as well as structural control on the distribution of ore, both within KCD, and other parts of the wider KZ Trend. Shear zones that were active during folding are a third key structural control on the location of ore within KCD and the wider KZ Trend. At KCD a folded carbonaceous shear in the core of the deposit juxtaposes stratigraphically distinct blocks, separating the 3000 lodes from the 5000 and 9000 lodes. The 3000 lodes above this shear are hosted by locally ferruginous cherts, carbonaceous argillites, and minor greywacke, whereas the 5000 and 9000 lodes below are hosted by siliciclastic rocks and banded iron formation. Fold shapes and wavelength differ between the two blocks reflecting their different rheologies during folding, and this is reflected in the scale, shape, and continuity of lodes in each block. At Pakaka and Kalimva, chlorite, carbonate, pyrrhotite, ± pyrite-altered shear zones rather than folds are the principal controls on gold distribution.

Sessenge (Sessenge-KCD Trend)

The Sessenge deposit is located approximately 1,000 m to the SW of the KCD deposit. Interpretations of drill data suggest the Sessenge deposit mineralisation represents the up-plunge

18^th September 2018 Page 67

Kibali Gold Mine NI 43-101 Technical Report

continuation of the KCD 9000 lode. Mineralisation at Sessenge forms one main mineralisation lode comprised of multiple small high-grade shoots that plunge to the NE towards KCD. Gold mineralisation is associated with silica-albite-siderite-ankerite-pyrite alteration, and occurs with an upper massive bedded ironstone unit, and a lower intercalated ironstone and clastic sedimentary unit. Strong chlorite alteration occurs on places, which has been erroneously logged as basic intrusive in historic RC drilling. The mean grade of the main low-grade halo is 1.5 g/t Au with several high-grade shoots between 4 g/t and 5 g/t Au.

Gorumbwa and Kombokolo Deposits (Sessenge-KCD Trend)

Both deposits occur along a NE trending mineralised corridor located some 800 m to the west of the main Sessenge-KCD structural zone, and each are considered to be part of the same mineralising event, with each having similar geologic, alteration and structural characteristics to the KCD deposit, however with mineralised zones of significantly smaller dimensions. High-grade shoots occur in the order of 30 m to 50 m as observed at KCD, but the surrounding low-grade mineralised halo occurring in the order of tens of metres, rather than tens to hundreds.

The Gorumbwa lode plunges at a low to moderate angle to the NE. In 1995, SOKIMO commenced mining from underground and small open pit operations. Total production is estimated at approximately 2.8 Mt at grades of approximately 7 g/t Au. The underground and open pit workings are presently collapsed and flooded though dewatering of the historic open pit is in progress to facilitate drilling programs. Two historic vertical shaft head frames remain to the east of the historic pit. Underground workings extend to 380 m below surface. The mineralisation consists of a series of stacked lenses that variably extend down plunge for a length of 1000 m at an average width of 200 m and has been identified at a depth of 400 m below topographic surface.

The lithological sequence based on mapping and core logging include a coarse meta sandstone sequence which is overlain by green schist facies metamorphosed meta conglomerate (coarse quartz clast) packages with intercalated medium to fine grained sediment horizons (meta arenite and meta siltstone units). The contact between meta sandstone sequence and upper coarse clastic sequences is marked by a thin matrix supported polymictic red chert or jasper clast bearing conglomerate horizon which is useful as a marker bed horizon. A dolerite unit sub parallel to lithological layering mapped in the SW is the main “Banc Vert”, another marker horizon that sits predominantly within the coarse meta sandstone unit immediately above or hanging wall of the main mineralised lens, the 1004 lens. From drill intersection interpretations the unit is interpreted to be sill-like. The Banc Vert is generally a medium grained mafic unit (dolerite) which becomes fine-grained at the contact with the host rock. The main constituents are chlorite, feldspar, and ankerite-calcite, with the ankerite appearing as porphyroblastic euhedral crystals up to 2 cm in diameter. Some parts of the mafic intrusion contain magnetite. The Banc Vert is not mineralised.

Mineralisation at Gorumbwa is hosted almost exclusively within the meta sandstone unit, with minor sporadic mineralisation noted in a conglomerate unit that occurs beneath the meta sandstone. Mineralisation is divided into twelve lenses like lodes (1001 to 1012) which coarsely trend west to west- SW, dip to the NW, and plunge to the ENE at approximately 30°. The lenses are echelon like in vertical stacking, with only the main 1004 lens being the most consistent in continuity. The stratigraphically higher upper lenses include 1001 to 1003, with 1005 to 1008 the

18^th September 2018 Page 68

Kibali Gold Mine NI 43-101 Technical Report

deeper footwall lenses. Higher grades within the lodes occur in the central area of the shoots where a higher strain environment increased space and hydrothermal fluid inflows. Historic mining focussed on the extraction of the main (1004) lode. The styles of mineralisation vary from KCD, with the dominant style being moderate to strong silicification and sericitisation with minimal pyrite. There is low correlation between sulphide and gold content. The second style is the typical ACSA style noted at KCD where the gold is proportional to pyrite percentages, though the iron carbonate is predominantly ankerite, unlike KCD which is dominated by siderite. This style is mainly observed in the main 1004 lode. The third style is visible gold within late, moderately to strongly silicification. Mineralisation is structurally controlled within a NE trending corridor where the S1 foliation strikes east-west within the central area rotating approximately 30° to 240° on the western edge of the corridor. The corridor is bounded by NE crosscutting structures on the eastern and western margins. These NE structures are indicated by the discontinuity of the red pebble conglomerate horizon in the west, micro folding within the lithological units near to the structures and rotation of S1 foliation laterally across the ± 200 m wide mineralised area. The structures may be ductile. Refolded folds are not observed but may have been seen in the high-grade shoot which was mined and makes up the void area.

The Kombokolo deposit lies approximately 1500 m to the NE of the Gorumbwa deposit, on the east side of an elongated topographic high locally named Kombokolo Hill. The hill is capped by an ironstone unit that strikes NE and dips gently-to-moderately to the NW. Mineralisation is located in a clastic conglomerate unit in the footwall of the ironstone, with moderate pervasively carbonate-sericite-silica-pyrite alteration. The mineralisation plunges at low to moderate angles (30°) to the NE. The lode has a down plunge continuation of over 300 m to a depth of 170 m below surface, an average width of 100 m, with an average thickness of 20 m. The lode remains open down plunge. The lowest stratigraphic unit at Kombokolo is a black to dark grey carbonaceous shale/argillite, which is overlain by a fine-grained, grey-coloured siltstone. The two sequences are overlain by the conglomerate.

Pakaka and Pamao Deposits (Pakaka-Mengu Trend)

The Pakaka-Pamao deposits are located at the SE end of the 7 km NW trending Pakaka-Mengu Trend. The stratigraphic section at Pakaka-Pamao is comprised of three formations; an upper tholeiitic basalt flow sequence with interbedded argillite and graphitic carbonaceous shale horizons called the Pakaka-Pamao Hanging Wall Formation; a middle sequence of met conglomerate interbedded with abundant felsic crystal tuff, undifferentiated tuff, as well as lesser horizons of siltstone and at Pamao, localised magnetite alteration. A lower footwall sequence of massive mafic volcanic units presently interpreted as a sequence of basalt flows is noted. The sequence could however be a relatively thick sill- or dyke-like intrusion.

Gold mineralisation at Pakaka-Pamao is hosted by the meta conglomerate interbedded with minor tuffaceous units. Recent works show mineralisation to be host in meta sandstone and banded iron formation. The mineralised zones are characterised by silica-ankerite-pyrite alteration, mainly in well foliated siliceous rocks. The mineralised zones are associated with pervasive silicification with local preservation of breccia textures that have been overprinted by the dominant S1 fabric. Higher gold grades appear to correlate well with the presence and abundance of pyrite, and empirically the high-grades appear to be spatially associated with the

18^th September 2018 Page 69

Kibali Gold Mine NI 43-101 Technical Report

intersection of the NW trending D1 thrust surface, and a NE trending strain corridor. The structures combine to produce a broad NE plunging open anticlinal structure, with Pamao on the west limb, and Pakaka on the east. The Pakaka mineralisation continues down plunge beyond the limits of the drilling and represents a further exploration potential. The Pakaka mineralisation extends over a strike length of 1,000 m, averages a thickness of 30 m and has been identified to a depth of 350 m below surface. The weathering profile at Pakaka is relatively deep.

Mengu Hill Deposit (Pakaka-Mengu Trend)

The Mengu Hill deposit lies near the NW end of the NW trending Pakaka-Mengu Trend. The stratigraphy in the vicinity of the deposit is dominated by a met conglomerate unit that is interbedded with fine-grained sediments, siliceous sericite schist and minor mafic volcanic rocks. These lithologies overlay a massive magnetite and specular hematite ironstone-chert unit that has weathered to create the topographic high – Mengu Hill – the ironstone protecting the northern face from weathering and erosion. Mineralisation is associated with silica-ankerite-pyrite alteration that is focused within the ironstone unit and along its contact with the overlying conglomerate unit. The mineralised lens is cigar like in shape and plunges shallowly to the NE with obvious development of refolded folds or sheaths. The Mengu Hill mineralisation averages a width of 150 m and continues 700 m down plunge to a depth of 250 m below the topographical surface.

Mengu Village and Marakeke Deposits (Pakaka-Mengu Trend)

At Mengu Village, located near the NW end of the Pakaka-Mengu Trend, the mineralisation is tabular in form, trending NW and dipping shallowly to the NE. The mineralisation is approximately 150 m in strike length with an average thickness of 15 m and has been identified to a depth of 150 m below the surface. The mineralisation is hosted by conglomerates with thin ironstone and carbonaceous shale intercalations.

The Marakeke deposit is located midway along the Pakaka-Mengu Trend with mineralisation developed in a variably carbonate-sericite-silica altered basalt and ironstone-chert, that dips to the NE at approximately 30° and strikes to the west-north-west. The Marakeke deposit occurs as a single tabular lens typically between 10 m to 30 m thick that trends NW and dips gently to the NE. The mineralised zone has a strike length of approximately 1,000 m and extends 200 m down dip.

18^th September 2018 Page 70

Kibali Gold Mine NI 43-101 Technical Report

8 Deposit Types

Gold deposits of the Kibali district are part of the globally significant group of Neoarchean orogenic gold deposits, examples of which are found in most Neoarchean cratons around the world.

Gold mineralisation within the Neo-Archean Kilo-Moto Belt is associated with epigenetic mesothermal style mineralisation, consistent with the majority of Archaean and Proterozoic greenstone terranes worldwide. The type of deposit has been termed orogenic gold and is generally associated with regionally metamorphosed terranes that have experienced a long history of thermal and deformational events and intrusion by igneous complexes. As such, the gold deposits are invariably structurally controlled. The most common style of mineralisation in this setting is fracture, vein-type and disseminated gold bearing sulphide mineralisation in zones of brittle fracture to ductile folding and dislocation.

The Kibali deposits differ from many orogenic gold deposits in terms of structural setting. Rather than being linked to a major large scale steeply dipping strike slip fault with brittle-ductile deformational evolution, they are hosted within a thrust stack sequence with ductile to brittle-ductile deformational structures and complex folding history.

The richly mineralised KZ Trend appears to have initiated as an extensional fault system along the boundary between the relatively young basin in the western part of the belt and older rocks to the east. Mineralisation occurred during the later stages of subsequent regional contractional deformation which resulted in inversion of the basin, development of reverse faults, and folds.

18^th September 2018 Page 71

Kibali Gold Mine NI 43-101 Technical Report

9 Exploration

The Project has been explored since the early 1990s by geochemistry sampling, mapping, trenching, geophysical surveys, and drilling. Kibali Goldmines has been exploring at Kibali since 2010. Exploration prior to Kibali Goldmines ownership is described in Section 6 of this Technical Report. During 2017, Kibali Goldmines spent approximately $7.3 M on greenfields and brownfields exploration.

9.1 Exploration Concept

The Kibali district is extremely prospective for gold mineralisation and the exploration approach is to locate suitable dilatational traps on, or adjacent to the main gold bearing structures that can host significant economic mineralisation.

Exploration at Kibali focusses on advancing both brownfields and greenfields targets. Brownfields exploration involves testing underground and open pit targets for extensions of high-grade mineralisation based on the structural model, but commonly in a down plunge direction as the major axis of continuity.

Satellite deposits and gaps between existing Mineral Resources are evaluated by exploration work to define Mineral Resources from conceptual targets. During 2018, a key exploration programme will target the previously identified prospect of Kalimva-Ikamva with the aim of defining Inferred Mineral Resources.

Recent geophysical surveys have been combined with a longer-term study to develop a tectonostratigraphy for Kibali, and to improve the understanding of the controls to gold mineralisation and regional geologic architecture. This Project -wide geologic framework is driving a re-assessment of exploration work to date as part of greenfields target generation.

9.2 Historical Exploration Review

All historical drilling completed by Barrick or Moto is reviewed on a case by case basis, prior to undertaking any key geological model reviews or updates of a Mineral Resource. Initially a number of historical holes are selected for twinning and this data is then used to make an informed decision as to the reliability of the historic data used for the Mineral Resource estimate.

In general, the twin holes completed to date have shown that assayed intercepts are mostly repeatable. However, some twin holes have identified that the ore intercept is at a different depth down hole relative to the historic data, thereby indicating that either the down hole survey or collar survey data of the historic data is not reliable.

In such instances, the historical holes have been removed from updated estimation datasets. However, there are still a small number of Inferred satellite resources which rely heavily on the

18^th September 2018 Page 72

Kibali Gold Mine NI 43-101 Technical Report

historic data namely Megi, Marakeke & Mengu Village, which currently poses a moderate level of risk to these Mineral Resources.

Historical maps have also been used in the definition of old underground mine workings particularly at Gorumbwa and Durba Hill. The Gorumbwa historical mine workings have been broadly confirmed with a phase of resource definition drilling during 2016 and 2017, which reduces the risk of a significant impact on the economic viability of the Mineral Resource. At Durba Hill, survey scanning, and probe drilling was completed in 2017.

9.3 Geophysics

Detailed interpretation of multi-source remote sensing datasets with ground checking of geologic and geophysical features forms the basis of Kibali Project exploration programs. Remote airborne data sets include magnetics and airborne electromagnetic (EM). The distribution and form of the ironstone units, carbonaceous shale horizons, and intrusive in the Project area are highlighted and in general can be mapped out by the airborne data sets. Targets with coincident magnetic highs (ironstone), EM conductive highs (carbonaceous shales), structural complexity with folding and dislocations, evidence of alteration and/or geochemical anomalism are of particular interest.

Spectrum Air Limited completed an airborne EM, magnetic and radiometric survey in 2010 over the Project (Figure 9-1 and Figure 9-2). A total of 10,559 line kilometres were surveyed at a nominal line spacing of 200 m, the KCD area was in filled to 100 m line spacing.

The airborne EM and magnetic data have both indirectly contributed to target generation by enhancing lithological and structural interpretations, and directly through detecting and outlining a number of NE plunging highly conductive linear shapes. Although the EM anomalies do not map actual gold mineralisation it is thought the conductive linear shapes highlight structurally prospective areas and have been interpreted as representing graphitic carbonaceous shale which has been deformed into a rod like shape by NE trending S2 structures. The magnetic anomalies delineate trends of ironstone units and highlight some of the intrusive bodies.

18^th September 2018 Page 73

18th September 2018 Page 74

18th September 2018 Page 75

Kibali Gold Mine NI 43-101 Technical Report

9.4 Geochemical Sampling

Soil samples are the first pass geochemical exploration technique used in the western portion of the licence area, where ease of access and suitable terrain aid field activities. Despite artisanal workings and potential surface contamination, the thin horizons of transported cover, shallow depths of paleo weathering surfaces (marked by quartz gravel layers in the district), and weak laterite development produce fairly robust geochemical anomalies which are in general proximal to sources of mineralisation. Samples are collected from B horizon levels, at 50 m centres along lines spaced 200 m and 400 m apart. Anomalous lines are in filled with samples at 50 m centres along lines spaced 100 m and 200 m apart. Soil samples are analysed using 50 g fire assay methods. Geochemical anomalies correlate well with NW trending D1 thrust surfaces (as evidenced by the Pakaka-Mengu trend) and NE trending S2 structure corridors.

In the eastern portion of the licence, thicker horizons of transported cover (>2 m) and higher grade metamorphism indicate further refinement of the interpretation of geochemical results is required. A review of multi-element and gold results in conjunction with one another highlights trends that can aid discrimination between real and transported anomalies. Numerous artisanal workings along the main Nzoro River tributary that transects the north and east of the licence area indicate that in certain catchment areas, a first pass stream sediment sampling program would be a more efficient geochemical approach, with ridge and spur, and grid based soil sampling a follow-up methodology.

Currently, geologic mapping, pitting, and trenching activities are completed prior to drill testing of geochemical and geophysical targets. Table 9-1 presents the Kibali Goldmines trenches, auger and pit exploration results that are in the Kibali database.

Table 9-1 Kibali Trenches, Auger and Pits Summary

Year	Company	Trenches		Auger		Pits		Total
		Meters	No.	Meters	No.	Meters	No.	Meters	No.
2010	Kibali Goldmines	481	5	-	-	273	48	754	53
2011	Kibali Goldmines	398	2	350	185	538	147	1,286	334
2012	Kibali Goldmines	1,050	43	1,083	181	691	131	2,823	355
2013	Kibali Goldmines	3,216	61	11	2	498	165	3,725	228
2014	Kibali Goldmines	8,570	83	83	23	1,115	383	9,768	489
2015	Kibali Goldmines	12,240	110	800	360	3,727	1,128	16,767	1,598
2016	Kibali Goldmines	8,066	101	1,799	843	1,830	648	11,694	1,592
2017	Kibali Goldmines	8,712	58	-	-	1,596	605	10,308	663
Total		42,733	463	4,126	1,594	10,268	3,255	57,125	5,312

18^th September 2018 Page 76

Kibali Gold Mine NI 43-101 Technical Report

9.5 Proposed 2018 Regional Exploration

Satellite deposits and gaps between existing Mineral Resources are evaluated by exploration work to define Mineral Resources from conceptual targets. During 2018, an exploration programme will target the previously identified prospect of Kalimva-Ikamva with the aim of defining Inferred Mineral Resources. Additional proposed regional exploration on the KZ structure during 2018 includes:

☐ Ngyoba – Drilling the 800 m gap potential between the Kibali River and Sessenge.

☐ KZ South – Zakito to Zambula and at Brindi.

9.6 Proposed 2018 Resource Definition Exploration

The 2018 resource definition exploration is scheduled to target the down plunge extension of the KCD 5000 lodes focussing above the bottom level of the shaft, with drilling from a dedicated underground exploration drill drive. Alongside continuation of the advanced grade control programme ahead of underground foot wall drive development and infill grade control programme for final pre-production definition of Measured Resources. Additional significant proposed resource definition drilling during 2018 includes:

☐ The up-plunge extension of the 3000 lode from the existing UG Ore Reserve area towards the KCD open pit.

☐ The up-plunge extension of the 9000 lode to connect to the Sessenge gap

☐ KCD open pit resource both 5000 and 3000 lodes will be further drill tested to potentially convert existing Inferred Mineral Resources to Indicated such that they can be incorporated into Ore Reserves.

☐ Drill test Mengu Hill down plunge extension to evaluate the underground potential.

9.7 Discussion

Kibali has a detailed Standard Operating Procedure (SOP) Manual for Exploration and Drilling Practices that provides standardisation and consistency for all field technical personnel to ensure the collection of quality data.

18^th September 2018 Page 77

Kibali Gold Mine NI 43-101 Technical Report

10 Drilling

10.1 Drill Hole Database

Table 10-1 presents the known drilling by year, company, and type, within the Kibali Permit.

Table 10-1 Kibali Drilling Summary

Year	Company	Diamond Drill		Reverse Circulation		RC Collar + DD Tail		Total
		Meters	No. of Holes	Meters	No. of Holes	Meters	No. of Holes	Meters	No. of Holes
1950	OKIMO	35,153	242	2,856	102			38,111	344
1951	OKIMO	1,259	15					1,259	15
1952	OKIMO	294	5					294	5
1960	OKIMO	16,162	175					16,162	175
1980	MOTO	1,484	10					1,484	10
1996	Barrick	8,986	70					8,986	70
2004	Moto	9,840	50	42,133	655			52,628	705
2005	Moto	42,672	201	52,229	749			95,650	950
2006	Moto	50,396	227	39,443	623			90,462	850
2007	Moto	51,404	125	21,830	412			73,646	537
2008	Moto	51,007	99	6,563	15			57,585	114
2009	Moto	23,035	67					23,035	67
2009	Kibali Goldmines	2,938	9					2,938	9
2010	Kibali Goldmines	28,403	64	29,427	494			58,324	558
2011	Kibali Goldmines	10,507	28	74,182	1,993			86,682	2,021
2012	Kibali Goldmines	23,493	88	100,228	1,876			125,597	1,964
2013	Kibali Goldmines	18,794	77	81,736	1,496			102,026	1,573
2014	Kibali Goldmines	34,079	176	152,636	3,181	595	4	189,896	3,952
2015	Kibali Goldmines	52,735	314	117,053	2,414	2,715	17	172,202	5,443
2016	Kibali Goldmines	71,435	572	214,215	2,973	8,691	48	288,623	12,236
2017	Kibali Goldmines	122,074	700	202,680	2,854			327,608	3,554
Total		656,150	3,314	1,137,211	19,837	12,001	69	1,813,197	35,152

The drilling history is summarised below:

☐ Over the course of several campaigns prior to 2009, a total of 459,301 m of historical drilling was conducted by previous operators as described in Section 6. The aim of this drilling was to estimate Indicated Mineral Resources as well as limited Measured Mineral Resources for feasibility study purposes.

☐ Since 2009, some 1,353,896 m of drilling has been. All Kibali Mineral Resources have been estimated using a combination of diamond drill (DD), reverse circulation (RC), and chip sampling.

☐ In addition to the table above, are a number of pseudo holes that have been created from trench/channel sampling. This information is, however, not generally used for resource modelling at Kibali (e.g. trenches for open pits and underground face sampling only for ore development).

18^th September 2018 Page 78

Kibali Gold Mine NI 43-101 Technical Report

10.2 Survey Grid

The Kibali mine uses the UTM Zone 35N datum WGS84 grid for drill hole coordinates.

10.3 Drill Planning and Site Preparation

Drill holes are planned in Vulcan and Micromine software. Consideration is given to the orientation of the drilling in relation to the geological structures, to provide for unbiased sampling.

The Senior Geologist, Drill Contractor, Mine Planner, Mine Surveyor, and Mineral Resource Manager all sign off on the drill hole plan prior to initiating drilling.

Open pit drill collars, as well as back sights and foresights, are surveyed in using hand held or differential Global Positioning System (GPS), and then staked, by the Kibali Mine Surveyors.

Underground drill collars, as well as back sights and foresights, are surveyed using total station underground survey instruments, and marked on the drift walls, by the Kibali Mine Surveyors.

There are three categories of drilling at Kibali:

☐ Exploration Drilling – wide spaced exploratory and resource definition drilling. This category will include diamond drilling with RC pre-collars (RC_DDH).

☐ Advanced Grade Control Drilling – consists of wider spaced drilling to position underground Footwall (FW) drives and also for Mineral Resource upgrades in the open pits.

☐ Infill Grade Control Drilling – used for final production definition to inform Measured Mineral Resources / Proved Ore Reserves. Generally, Kibali’s inventory of infill grade control drilling is some three to six months inventory for open pit and approximately 18 months for underground.

10.4 Downhole Surveying

Reflex EZ-Trac tools were used prior to mid-2016 but were replaced by Reflex EZ-Gyro. When both EZ-Trac and conventional Gyro surveys were being completed, the results of the Gyro survey took higher priority than those of Reflex EZ-TRAC surveys.

All drill holes are now surveyed down hole with a Reflex EZ-Gyro.

Downhole survey equipment is calibrated yearly and checked every quarter by Reflex technicians during site visits.

10.5 Collar Surveys

All drill holes have had their collar location surveyed using differential GPS to a 10 mm accuracy.

18^th September 2018 Page 79

Kibali Gold Mine NI 43-101 Technical Report

10.6 Diamond Drilling

DD is utilised for resource extension work. PQ rods (85.0 mm) are generally used for the first 100 m down hole with HQ (63.3 mm) or NQ (47.6 mm) used from 100 m to 200 m depending on the drilling depth requirement. All grade control diamond drilling is completed in NQ.

The diamond drilling has been completed by Boart Longyear (surface DD) and Ore Zone (underground DD), dominantly in NQ size.

Core recoveries are in general good, with an average of 98.8% recovery in the unweathered rock, 94.3% recovery in the transitional zone and 73.6% in saprolite zone. Average ore zone recovery was 98.7% with a range of between 70% and 100%.

Drilling Procedure

A Project geologist must be on site prior to drilling commencing and ensure that the drill rig is lined up as per the drill plan as well as supervising drilling, core orientation and down hole surveying. Once each drilling run is complete, the drill core is removed from the drill rod and placed in an angle iron rack to mark up an orientation line with red chinagraph pencil or crayon from the data received from a Reflex ACT II Core Orientation Tool. The apex of the structure is marked on the core in a chinagraph pencil or crayon by the core technician. If the orientation and apex lines are overlapping, then the apex line is offset by 5 mm.

Diamond core is transferred to the core trays and a plastic down hole depth marker is placed at the beginning and end of each core run with the depth marked on it. All areas of core loss are identified and the core is marked up for core recovery. Each drill core box is marked with the Hole ID, top and bottom depth of the core and the box number. The core is then transferred to the core yard facility by Kibali Goldmines staff for logging and sampling.

Core Logging

Diamond drill core is geologically logged and includes weathering, mineralisation, alteration, lithology, structure and redox. This is stored in a central database after validation. The core is also digitally photographed.

All diamond drill core is oriented and where orientation is not possible the core is assembled with previous runs, where possible, in an attempt to extend the orientation line.

Geotechnical logging is only performed for holes drilled specifically for geotechnical assessment as required.

Logging is completed on hard copy before being transcribed to the database. Digital logging systems are due to be implemented during 2018.

18^th September 2018 Page 80

Kibali Gold Mine NI 43-101 Technical Report

Sampling

Diamond drill half core samples are taken within geological units and are normally between 0.8 m and 1.2 m long. The drill core is split using diamond saws utilising fresh water. Half core is submitted for assay wherever possible, quarter core is only used when there is a need for confirmation purposes on the same interval.

Diamond core is photographed, both wet and dry, before being halved with a diamond saw. One half is submitted for assay analysis whilst the other half is stored for future reference.

10.7 Reverse Circulation Drilling

RC chip samples are logged with the same lithological, mineralogical and alteration information as DD core but are logged on the 2 m RC samples from the riffle splitter.

Drilling Procedures

RC holes are used for advanced grade control and infill grade control drilling using 131 mm diameter rods. RC chips are sieved and logged by the site geologist before being placed in chip boxes for storage.

Sampling and Splitting Procedure

RC samples are collected from the rig in two metre intervals using a riffle splitter. Auxiliary booster units are used to ensure the vast majority of the samples collected are already dry. On the rare occasion a wet sample is obtained it is dried before being manually split.

Reverse Circulation Logging

Reverse circulation (RC) samples are riffle split and composited on 2 m downhole samples. RC drilling is completed by Boart Longyear and Ore Zone.

RC sample recovery is measured by weighing the total weight of sample collected over a meter drilled and comparing it to the theoretical expected weight for each material type (lithological unit) and weathering type.

Sample size optimisation has been completed for each deposit to verify the suitability of the sample size.

10.8 Drill Twinning Studies

Twin drilling studies, originally undertaken at Kibali as part of the original feasibility studies prior to construction of underground mine development, continue as part of the Mineral Resources Management programme. These comparisons have shown that although there can be variations in grade, as expected, the broad intercepts and relative grade of the intersections are comparable across the twins.

18^th September 2018 Page 81

Kibali Gold Mine NI 43-101 Technical Report

The variation in the drill twinning studies has been used to feed the classification criteria for the Mineral Resources.

10.9 Kibali Mineral Resource Drill Spacing Optimisation

Drilling directions for Mineral Resources namely Pakaka, Kombokolo, Sessenge, Pamao, Gorumbwa, Mengu Hill, and KCD open pit are optimised on an individual deposit basis to ensure that the preferred drilling direction for Resource and Grade Control, drilling is on a cross plunge basis.

Measured classification grade control drill spacing has been independently optimised using change of support analysis. In general, the infill drill spacings ranges between 10 m to 20 m along the principal direction and 5 m to 10 m across strike within the ore zones and are sampled at 2 m downhole intervals.

Indicated classification grade control drill spacing has been independently optimised using change of support analysis. In general, they are spaced approximately 40 m by 40 m with geological continuity of 100 m and more along strike. All open pit resources that also form reserves, namely KCD, Kombokolo, Pakaka, Pamao, Gorumbwa, and Sessenge have been drilled to an advanced grade control spacing.

Inferred classification Resource drill holes are on an 80 m by 80 m or less drill spacing.

The data distribution is one of several classification specifications for the resource estimate which includes minimum sample points, kriging variance and conditional bias.

All drilled holes are composited to 2 m down hole during resource estimation; this is supported by sample interval optimisation study completed that show 2 m is optimal for sampling within Kibali Permit.

10.10 Other Sampling Methods

Other sampling methods, such as grab sample, channel samples and soil samples are also used based on the area of interest and stage of exploration.

Chip samples are used within the underground development area to provide an additional source of information regarding the mineralisation associated with the alteration, particularly when mapping low-grade halo contacts. This data is recorded on the underground geological maps, which are then scanned and georeferenced for wireframe model updating. However, this data is not used for estimation.

Rotary air blast drilling (RAB) drilling is used in regional first pass exploration and for sterilisation purposes. No samples used for Mineral Resource estimation are from RAB holes.

18^th September 2018 Page 82

Kibali Gold Mine NI 43-101 Technical Report

Kibali geologists also conduct geological and structural mapping in all accessible open pit and underground development areas. This data is scanned and georeferenced so that it can be utilised in Mineral Resource modelling.

10.11 Discussion

In the QP’s opinion, the drilling and sampling procedures at Kibali are robust, suitable for the style of mineralisation and are at or above industry standard practices. There are no drilling, sampling, or recovery factors that could materially impact the accuracy and reliability of the results.

18^th September 2018 Page 83

Kibali Gold Mine NI 43-101 Technical Report

11 Sample Preparation, Analyses and Security

11.1 Sample Selection

The sample boundaries of drill core are determined based on geology and alteration and most often varies from 0.8 m to 2.4 m. Half core is used whenever possible. Historically, quarter core was only used when there is a requirement for a duplicate assay or if another analysis type was required for the same interval. Metallurgical samples are taken from dedicated metallurgical drill holes or from coarse reject material.

RC samples are collected from the rig in two metre intervals using a riffle splitter to create a 3 kg to 4 kg sample. Wet samples are dried before being split.

11.2 Sample Preparation

All samples submitted for assay are prepared and analysed at SGS Doko laboratory, which is managed and self-certified by SGS and located on the Kibali mine site.

Grade control and exploration drill samples are prepared in the same manner. Once the samples are received by SGS Doko, the sample is weighted and entered into a LIMS tracking system. Samples are dried in an oven at 105°C. Channel and trench samples are disaggregated to remove dry lumps. Dried samples are crushed to ensure that 75% of the sample is below 2 mm.

The crushed sample is then passed through a BOYD splitter and the reject material is retained. The split sample is then pulverised in an LM2 pulveriser until 85% passes through a 75-micron (200 mesh) screen and a 350 g is split removed and placed in a packet. The LM2 pulveriser is cleaned with an air hose every sample, and with blank material every 6th sample. SGS Doko undertakes regular screen sieve tests on the crushing and pulverising. The coarse (2 mm) reject and the pulp (75 micron) reject material are returned to Kibali for storage at the mine site and future re-analysis if required.

An external audit completed by Optiro in 2017 concluded that the sample preparation procedures followed standard industry practices.

The QP concludes that the sample preparation procedures are regularly checked and follow industry standard practices.

18^th September 2018 Page 84

Kibali Gold Mine NI 43-101 Technical Report

Figure 11-1 outlines the preparation and analysis flow chart for diamond drill core samples.

Figure 11-1 Diamond Drill Core Sample Flowchart

18^th September 2018 Page 85

Kibali Gold Mine NI 43-101 Technical Report

Figure 11-2 outlines the preparation and analysis flow chart for RC samples

Figure 11-2 Reverse Circulation Sample Flowchart

18^th September 2018 Page 86

Kibali Gold Mine NI 43-101 Technical Report

Figure 11-3 outlines the preparation and analysis flow chart for RC samples

Figure 11-3 Channel Sample Flowchart

18^th September 2018 Page 87

Kibali Gold Mine NI 43-101 Technical Report

11.3 Sample Analysis

All samples for Kibali are analysed by SGS Doko laboratory at the Kibali mine site or SGS Mwanza laboratory in Tanzania. SGS Mwanza is used for sample overflow and analysis that could not be completed at SGS Doko including multi element, arsenic for selected samples and soils analysis. Both laboratories are operated independently and self-certified by SGS.

All samples are analysed using lead collection 50 g fire assay with atomic absorption finish with a gravimetric finish for any samples reporting above 100 g/t Au.

Results discussed include samples from brownfield exploration and resource, open pit grade control, and underground grade control. A total of 249,359 samples were submitted in 2017. Approximately 15% of the total samples received are check samples inserted into the sample streams (Table 11-1). Check samples consist of field duplicates for RC, pulp duplicates for diamond cores, certified referenced materials (CRM) and coarse blanks.

Table 11-1 2017 Submitted Samples

Sample Type	Number of Samples	Percentage
DDH	105,467	42%
RC	102,176	41%
Others	4,656	2%
Subtotal	212,299	85%
Standards	12,497	5%
Blanks	12,140	5%
Duplicates	12,423	5%
Subtotal	37,060	15%
Total	249,359	100%

11.4 Quality Assurance and Quality Control

Kibali has an extensive Quality Assurance and Quality Control (QA/QC) programme in place. This section covers the QA/QC from 1^st January 2017 to 20^th December 2017 (the review period) for brownfields exploration and grade control assay data. Previous QA/QC reporting periods have not been observed to contain any significant sources of error or bias which would have a material effect on the Mineral Resource.

Quality Assurance (QA) is to demonstrate that the sampling and analytical protocols are appropriate and optimal for the deposit in question. It should entail orientation sampling studies and statistical analysis so that appropriate systems and standards can be tailored to achieve quality results throughout all the stages of collecting and analysing data. Ideally, orientation studies are performed at the beginning of or early stages of project evaluation. Setting up systems and standards to ensure quality throughout all of the stages used to collect and analyse data

Quality Control (QC) is a real-time monitoring and analysis to ensure the protocols developed in QA are being adhered to and are returning precise and accurate results. Entails additional sampling and analysis and statistical examination (such as scatter plots, quantile-quantile (QQ) plots etc.).

18^th September 2018 Page 88

Kibali Gold Mine NI 43-101 Technical Report

Quality control checks are inserted into sample stream prior to dispatch to the laboratory except for diamond core duplicates which are taken as a split by Kibali Goldmines staff in the laboratory using a riffle splitter after crushing but before pulverising. Overall, the QA/QC sampling includes 5% field duplicates, 5% blanks, and 5 % CRM. Umpire (secondary) independent laboratories are also used on quarterly basis as check for the primary laboratory, as well as to check the consistency in sampling protocols.

During 2017, 12,497 CRMs, 12,140 blanks, and 12,423 field duplicates were submitted with the field samples.

All laboratories undertake their own internal QA/QC which includes blanks, duplicates and CRMs, which are reported to Kibali along side the field sample results. These results of the laboratory internal QA/QC is reviewed separately by Kibali, but are not reported below.

The Kibali QA/QC protocol flowchart illustrated in Figure 11-4.

Certified Reference Material

Certified reference materials (CRMs) are inserted into batches at a frequency of 1 in 20 (5%) samples to check for bias over time and to test for laboratory handling errors. These monitor the accuracy of results received from the laboratory by comparing it with the certified reference value.

All CRMs used in the review period are sourced from Ore Research and Exploration Pty Ltd, Australia, and are oxide or sulphide type with a matrix of feldspar minerals, basalt, and iron pyrites. CRMs are purchased in pre-packaged 50 g samples that require no preparation before being submitted to the laboratory. A sub-set of the total CRMs available are used and are rotated on a quarterly basis to prevent laboratory identification.

CRM results are monitored and classified as a failure if one sample point falls outside of three standard deviations from the certified mean, or three consecutive samples fall outside of two standard deviations (on the same side) of the mean.

CRM results that have a failure outside of three standard deviations are checked for possible CRM swaps. This is investigated by comparing the returned assay grade to the list of known CRM grades values. The CRM samples are supplied by OREAS with CRM ID printed on the bag. This printed ID is photographed during CRM insertion and then removed prior to submission of the CRM to the laboratory. This CRM photograph is used to help identify CRM swaps. A normal sample swap is also investigated to check if a normal drill sample has been labelled as CRM.

18^th September 2018 Page 89

Kibali Gold Mine NI 43-101 Technical Report

Figure 11-4 Kibali QA/QC Protocol Flowchart

18^th September 2018 Page 90

Kibali Gold Mine NI 43-101 Technical Report

In addition to the CRM photographs, swaps can be investigated using the technician’s sampling plan document, verifying used sample numbers, reviewing the sample booklet, and comparing against the other CRMs in the batch When all the above investigations are complete, and it has been established that a failure has occurred, the following actions are initiated:

☐ When two or more CRMs failed in batch and the failure is as result of sample swap, the entire batch is called for re-assay.

☐ When one or more CRM failed in batch and the failure is not as result of sample swap, the entire batch is called for re-assay.

Based on the above controls when a batch is re-assayed and fails again, the samples are flagged but committed into the database whilst new samples are prepared for re-analysis. If a CRM is observed as repeatedly failed over a period of time, then it is removed from storage and is no longer inserted into the samples stream.

SGS Doko

A total of 10,004 commercial standards have been submitted to SGS Doko during the reporting period. Table 11-1 lists all CRMs analysed at SGS Doko during this period.

Overall performance across all standards shows that 97% of the samples passed within 10% expected values. Remaining 3% of the total data analysed differs by more than 10% from the expected value but are within the 20% acceptable limit however, most of these samples have been re-assayed and few still returned poor results.

Investigations indicate that most of these failures are potentially the results of CRM OREAS250 that has been mostly contributing to most of the failures. The CRM is of low-grade (0.31 g/t Au) and due to homogeneity problem, it returned failed values from both SGS Doko and Mwanza. These have been removed from the list of in use CRMs and are no longer inserted in sample stream.

Table 11-2, Table 11-3 and Table 11-4, as well as Figure 11-5 summarise the performance of the CRMs relative to their upper and lower limits plotted on a scatter plot.

18^th September 2018 Page 91

Kibali Gold Mine NI 43-101 Technical Report

Table 11-2 List of CRMs Assayed from SGS Doko

ID	Cert Value (g/t Au)	CertSt dDev	Min Cert (g/t Au)	Max Cert (g/t Au)	Min Assay (g/t Au)	Max Assay (g/t Au)	Mean Assay (g/t Au)	St Dev	Sample Count
OREA   S 250	0.309	0.013	0.27	0.348	0.26	0.42	0.30	0.02 7	466
G310-6	0.65	0.04	0.57	0.73	0.49	0.78	0.65	0.03	2,545
G910-   10	0.97	0.04	0.89	1.05	0.85	1.14	0.96	0.03	1,234
OREA   S 205	1.24	0.05	1.09	1.39	1.11	1.39	1.26	0.05	2,974
OREA   S 254	2.55	0.076	2.32	2.78	2.19	2.93	2.55	0.08	1,047
G314-5	5.234	0.21	5.024	5.864	5.1	5.62	5.33	0.10	149
OREA   S 210	5.49	0.15	5.19	5.79	5.04	5.72	5.41	0.11	491
G909-4	7.52	0.3	6.92	8.12	6.63	7.78	7.47	0.11	547
G307-7	7.75	0.445	7.305	9.085	7.42	7.94	7.74	0.22	15
OREA   S 208	9.25	0.44	8.81	10.57	8.43	10.3	9.38	0.4	546
Total									10,004

Table 11-3 CRM Summary for Review Period at SGS Doko

STD_ ID	Minimum Assay     (g/t Au)	Maximum Assay     (g/t Au)	No of     Samples	%Pass of - /+1STD	%Pass of - /+2STD	%Pass of - /+3STD
All CRM	0.31	10.30	10,004	82%	97%	99.7%

Table 11-4 CRM Statistics for SGS Doko

Statistics	Expected	Assayed	Units	Distribution	Expected	Assayed
Population	10,004			25.0%	0.65	0.67
Minimum (g/t)	0.31	0.26	g/t	50.0%	0.65	0.74
Maximum (g/t)	9.25	10.30	g/t	75.0%	0.97	0.99
Mean (g/t)	2.21	2.21	g/t	80.0%	1.24	1.24
Std Dev (g/t)	2.49	2.51	g/t	90.0%	1.24	1.28
CV	1.13	1.13		97.5%	1.24	1.31
Correlation	0.999			99.9%	2.55	2.50

18^th September 2018 Page 92

Kibali Gold Mine NI 43-101 Technical Report

Figure 11-5 Scatter Plot of CRMs at SGS Doko

Though there are some failures noted during the period, one of the low-grade OREAS CRM (OREAS250) failed repeatedly. After analysing & investigating it at both SGS Doko and Mwanza, it was decided not to use this CRM furthermore. Samples were committed to database after all investigations have been completed and sources of failure identified like CRM homogeneity and mismatch with geology. Tramline analysis is used to identify possible CRM sample swaps in samples inserted into the sample stream (Figure 11-6). This has been grouped into low-grade, medium-grade and high-grade ranges.

18^th September 2018 Page 93

Kibali Gold Mine NI 43-101 Technical Report

Figure 11-6 Tram Line Graph for CRMs Analysed at SGS Doko

18^th September 2018 Page 94

Kibali Gold Mine NI 43-101 Technical Report

SGS Mwanza

A total of 2,329 commercial standards have been submitted to SGS Mwanza during the review period. Table 11-5 lists all CRMs assayed at SGS Mwanza.

Table 11-5 List of CRMs Assayed at SGS Mwanza

ID	Cert   Value   (g/t Au)	CertStd   Dev	Min   Cert (g/t   Au)	Max   Cert (g/t   Au)	Min   Assay   (g/t Au)	Max   Assay   (g/t Au)	Mean   Assay   (g/t Au)	St   Dev	Sample   Count
G310-6	0.65	0.04	0.55	0.77	0.6	0.7	0.65	0.03	520
G910-10	0.97	0.04	0.89	1.05	0.91	1.06	0.98	0.036	160
OREAS20   5	1.24	0.05	1.09	1.39	1.11	1.33	1.25	0.011	973
OREAS25   4	2.55	0.076	2.322	2.778	2.39	2.69	2.54	0.057	108
OREAS21   0	5.49	0.15	5.04	5.94	5.24	5.38	5.27	0.027	5
G909-4	7.52	0.3	6.62	8.42	6.86	8.74	7.41	0.027	144
G307-7	7.75	0.445	6.415	9.085	7.74	8.32	8.08	0.24	11
OREAS20   8	9.25	0.44	7.93	10.57	8.86	10.4	9.17	0.4	408
Total									2,329

Overall performance on all CRMs show that 99.6% of CRM samples pass within two times standard deviation. Of the total data analysed, 0.4% differ by more than 10% from the expected value but within 20% acceptable limit.

In total, less than 0.4% of total samples reported outside of the 2nd standard deviation. Most of these have been re-assayed and re-reported well within 2nd Standard Deviation (Table 11-6 and Table 11-7). There were few CRM swaps identified that were corrected in the database.

Table 11-6 CRM Summary for SGS Mwanza

STD_ID	Minimum Assay	Maximum Assay	No of Samples	%Pass of- /+1STD	%Pass of -- /+2STD	%Pass of- /+3STD
All CRM	0.31	10.40	2,329	86%	99.6%	100%

Table 11-7 CRM Statistics for SGS Mwanza

Statistics	Expected	Assayed	Distribution	Expected	Assayed
Population	2,329		25.0%	0.97	0.97
Minimum (g/t)	0.31	0.30	50.0%	1.24	1.20
Maximum (g/t)	9.25	10.40	75.0%	1.24	1.24
Mean (g/t)	2.98	2.97	80.0%	1.24	1.25
Std Dev (g/t)	3.33	3.29	90.0%	1.24	1.26
CV	1.11	1.11	97.5%	1.24	1.29
Correlation	1.000		99.9%	2.55	2.63

Figure 11-18 illustrates the performance of the CRM’s relative to their upper and lower limits plotted on a scatter plot. Specific standards have been noted to be mostly contributing to the most of the failure. They have been removed from the sample stream and tested at other laboratories.

18^th September 2018 Page 95

Kibali Gold Mine NI 43-101 Technical Report

Though there are some level of failures noted in the year, continual work to improve the overall sampling protocol has been reviewed and implemented. Samples were committed to database after all investigations had been completed and error sources identified.

Figure 11-7 Scatter Plot for CRMs Used Between January and December 2017

A tramline is used to identify possible swaps in samples inserted into the sample stream. This has been grouped into low-grade, medium-grade and high-grade range (Figure 11-8).

18^th September 2018 Page 96

Kibali Gold Mine NI 43-101 Technical Report

Figure 11-8 Tram Line Graph for CRMs Analysed at SGS Mwanza During the Review Period

18^th September 2018 Page 97

Kibali Gold Mine NI 43-101 Technical Report

Blanks

Blank samples are assayed to help ensure no false-positives are obtained from the laboratories and to check for sample contamination. These samples should return gold assay values below the analytical detection limit (i.e. <0.01 g/t Au). The coarse blank samples used on the Project for this study are prepared onsite from barren granite material sourced from Matiko and Kalimva, about 20 km NW of the Project area. In 2018, blank sample raw material will be replaced with OREAS certified blanks that reflect the geology on site.

During the collection of samples, blank sample materials were inserted into sample stream at a rate of about 1 in 20 (5%) of the total submitted samples. These samples undergo the same sample preparation as the drill samples and used to detect inter-contamination due to poor cleaning of sample preparation equipment throughout the various sub sampling process.

SGS Doko

A total of 5,999 blank samples have been submitted to SGS Doko. The results are evaluated against twice the standard deviation as an acceptable limit. The overall performance shows more than 99.7% of the blanks samples assayed fell within the 10% acceptable limit (Table 11-8 and Figure 11-9). Overall the performance is deemed as good.

Table 11-8 Statistics for Blank Samples at SGS Doko

STD_ID	Minimum Assay	Maximum Assay	Mean Assay	Assayed StdDev	No of Sample	Expected Value	CRM StdDev	%Pass of - /+1STD	%Pass of - /+2STD	%Pass of - /+3STD
BLANK	0.005	0.22	0.01	0.01	9,861	0.03	0.02	52.3%	99.7%	99.8%

Figure 11-9 Performance Graph of Blank Samples at SGS Doko During the Review Period

18^th September 2018 Page 98

Kibali Gold Mine NI 43-101 Technical Report

SGS Mwanza

A total of 2,279 blanks have been submitted to SGS Mwanza. Again, the results have been analysed against twice the standard deviation as an acceptable limit. The overall performance shows more than 99.9% of the blanks assayed at SGS Mwanza is within 10% acceptable limit (Figure 11-10 and Table 11-9). The overall performance of the blanks is deemed good.

Table 11-9 Statistics for Blank Samples at SGS Mwanza

STD_ID	Minimum Assay	Maximum Assay	Mean Assay	Assayed StdDev	No Sample	Expected   Value	CRM StdDev	%Pass of - /+1STD	%Pass of   -/+2STD	%Pass of   -/+3STD
BLANK	0.005	0.1	0.01	0.01	2,279	0.03	0.02	71.5%	99.9%	100.0%

Figure 11-10 Performance Graph for Blank Samples at SGS Mwanza

Duplicates

Duplicate samples are used to check the homogeneity of the samples prior to sample splitting along with the accuracy and repeatability of sampling, splitting and assaying A duplicate sample is inserted after every 20^th sample.

Duplicate samples can be obtained from three sources

☐ Field Duplicates are obtained from the initial splitting of the RC sample during sampling at the rig

☐ Coarse (Reject) Duplicates are obtained from the coarse reject sample that is returned from the laboratory after the initial crush to 6 mm of the entire RC or half core sample.

18^th September 2018 Page 99

Kibali Gold Mine NI 43-101 Technical Report

☐ Pulp Duplicates are obtained from the pulverised 75 micron sample that is returned from the laboratory after the pulp is removed for analysis.

Field duplicates are not undertaken on diamond drill samples as the variance between the two half of core, due to the nuggety nature of gold mineralisation.

To more accurately quantify and address the sources of bias seen in duplicates, RC and DD samples have been reviewed independently.

Insufficient quantities of pulp duplicates were analysed at SGS Doko or SGS Mwanza during the review period. The umpire samples have been reviewed as a proxy. Kibali plan to submit pulp duplicates at a rate of 1 in 20 samples from 2018 to match the other duplicate insertion rates. No significant bias was observed in the pulp duplicates that were analysed.

RC Field Duplicates

A total of 6,446 RC field duplicate samples were analysed from Kibali at SGS Doko and SGS Mwanza within the period under review.

If a batch is suspected of having a high variation in duplicate values then the sampling protocol is checked, and the laboratory is asked to re-assay a range, or indeed the whole batch, depending on the confidence in the initial results.

SGS Doko

A total of RC field 6,446 duplicates samples was analysed at SGS Doko during 2017 (Table 11-10).

Table 11-10 Statistics for RC Duplicates at SGS Doko

Statistic	Original	Duplicate	Distribution	Origin	Duplicate
Population	6,446		25.0%	0.02	0.02
Minimum (g/t)	0.01	0.01	50.0%	0.02	0.02
Maximum (g/t)	68.50	65.80	75.0%	0.04	0.03
Mean (g/t)	0.60	0.60	80.0%	0.06	0.06
Std Dev (g/t)	1.99	1.94	90.0%	0.12	0.12
CV	3.30	3.23	97.5%	0.26	0.27
Correlation	0.996		99.9%	0.41	0.40

The data suggests there is no significant bias between the original samples and field duplicates. Overall pair data is acceptable, but this could be improved with implementation of more stringent measures. Figure 11-11 illustrates a QQ plot of the original versus field duplicate results. This plot indicates a good correlation between the original and field duplicate samples. Duplicates are noted to have a slight lower variance above 10 g/t Au but with 10% limit and does not have any material effect on the confidence in the resource estimate as these samples represent less than 0.5% of the total samples assayed.

18^th September 2018 Page 100

Kibali Gold Mine NI 43-101 Technical Report

Scatter plots are used in the QA/QC programme at Kibali to identify point to point sample relation, thereby helping to identify areas where there is significant grade difference. Overall correlation observed in Figure 11-11 is good and most results lay within the 10% acceptable range.

Precision plots, also known as relative difference plots, (Figure 11-12) shows that the 79% of RC field duplicates analysed at SGS Doko had sample pairs within a 20% difference of each other and 63% of these samples had samples pairs within 10% of each other, which is considered an acceptable level of variance given the natural variability in the deposits. Usually 80% of data within 20% difference with a threshold of 64% of pairs within 10% difference is deemed acceptable. The overall performance at SGS Doko is acceptable but is slightly below the expected 10% and 20% threshold.

The performance of duplicates needs to be improved further by implementing more rigorous sampling and splitting techniques. More stress is being given to sample splitting techniques. New sample splitters (Gilson Splitter) recommended by Optiro are also implemented at all the RC rig sites. The Gilson splitter is of larger capacity and relatively easy in obtaining an unbiased split sample.

Figure 11-11 Ascending Grade Correlation Plot for RC Samples Between 0 g/t Au to 40 g/t Au at SGS Doko

18^th September 2018 Page 101

Kibali Gold Mine NI 43-101 Technical Report

Figure 11-12 Precision Plot of RC Field Duplicates vs Original Sample at SGS Doko

Coarse Duplicates

Coarse duplicates referenced in this report are duplicates that have been split at a SGS laboratory between initial crushing and sample pulverising. A total of 4,171 coarse duplicates were analysed at SGS Doko and SGS Mwanza during the reporting period.

SGS Doko

Table 11-11 summarises the statistical analysis of the coarse duplicates submitted to SGS Doko. The data suggests that there is insignificant bias between the original samples and duplicates. Overall pair data is acceptable but can still be improved with protocols.

Table 11-11 Statistics of Coarse Duplicates at SGS Doko

Statistic	Original	Duplicate	Distribution	Original	Duplicate
Population	4,171		25.0%	0.02	0.02
Minimum (g/t)	0.01	0.01	50.0%	0.03	0.03
Maximum (g/t)	89.90	80.10	75.0%	0.05	0.04
Mean (g/t)	1.56	1.55	80.0%	0.08	0.08
Std Dev	4.60	4.54	90.0%	0.17	0.17
CV	2.95	2.94	97.5%	0.48	0.48
Correlation	0.997		99.9%	0.84	0.83

The correlation plot presented in Figure 11-13 shows a good correlation with an evenly distributed grade in both original and duplicates. Duplicates grades are noted to be slightly lower between 12 g/t Au and 15 g/t Au but within 10% threshold. However, this does not pose a major concern to the confidence in the resource as these samples are less than 0.5% of the total samples

18^th September 2018 Page 102

Kibali Gold Mine NI 43-101 Technical Report

assayed. This variation is expected in this type of gold deposit where the nugget effect is within 15% to 30% between samples.

Figure 11-13 Ascending Grade Correlation Plot for SGS Doko Coarse Duplicates Between 0 g/t Au to 50 g/t Au

From scatter plots analysis (Figure 11-14) a good correlation is observed with most samples falling within a 10% acceptable range. The data presents a correlation coefficient of 0.997 which is deemed acceptable.

Precision plots (Figure 11-15) shows that 84% of the coarse duplicates analysed at SGS Doko returned paired results within a 20% difference of each other and 68% of the coarse duplicates within 10% difference. Usually 80% of data within 20% difference with a threshold of 64% of pairs within 10% difference is deemed acceptable. The overall performance is in line with the expected threshold.

18^th September 2018 Page 103

Kibali Gold Mine NI 43-101 Technical Report

Figure 11-14 Normal Scatter Plot of SGS Doko Coarse Duplicates £ 50 g/t Au Tail Cut

Figure 11-15 Precision Plot of Coarse Duplicates vs Original Sample at SGS Doko

SGS Mwanza

A summary of the coarse duplicates submitted to SGS Mwanza is presented in Table 11-12.

18^th September 2018 Page 104

Kibali Gold Mine NI 43-101 Technical Report

Table 11-12 Statistics for Coarse Duplicates at SGS Mwanza

Statistic	Original	Duplicate	Distribution	Original	Duplicate
Population	2,145		25.0%	0.03	0.03
Minimum (g/t)	0.01	0.01	50.0%	0.04	0.04
Maximum (g/t)	170.00	169.00	75.0%	0.06	0.06
Mean (g/t)	2.14	2.14	80.0%	0.11	0.11
Std Dev	7.57	7.51	90.0%	0.20	0.20
CV	3.54	3.51	97.5%	0.63	0.65
Correlation	0.999		99.9%	1.15	1.13

The data indicates a significant correlation between the original and duplicates samples. A correlation of 0.999 indicates a low variation. The QQ plot displayed Figure 11-16 shows a very good correlation with an evenly distributed grade between both original and duplicate samples.

Figure 11-16 Ascending Grade Correlation Plot for Mwanza Coarse Duplicates from 0 g/t Au to 60 g/t Au at SGS Mwanza

From the scatter plot analysis (Figure 11-17) an overall very good correlation can be observed with most samples falling within the acceptable range of 10%. The precision plot (Figure 11-18) shows that 95% of the coarse duplicates analysed at SGS Mwanza returned paired results within a 20% difference of each other. A threshold of 80% of the data falling within a 20% difference with 83% of pairs within 10% is deemed acceptable. Overall the performance is very close to the expected threshold.

18^th September 2018 Page 105

Kibali Gold Mine NI 43-101 Technical Report

Figure 11-17 Normal Scatter Plot of Duplicates for Data at 60 g/t Au Tail Cut at SGS Mwanza

Figure 11-18 Precision Plot of Pulp Duplicates vs Original Sample at SGS Mwanza

18^th September 2018 Page 106

Kibali Gold Mine NI 43-101 Technical Report

Field Trench (Channel) Duplicates

Field channel sample duplicates references in this report are field duplicates that have been split at the laboratory after splitting trench channel sample. A total of 271 trench field duplicates were analysed at SGS Doko during the reporting period.

SGS Doko

Table 11-13Error! Reference source not found. summarises the statistical analysis of the Trench duplicates submitted to SGS Doko. The data suggests that there is insignificant bias between the original samples and duplicates. Overall pair data is acceptable but can still be improved with protocols.

Table 11-13 Statistics of Trench Duplicates at SGS Doko

Statistic	Original	Duplicate	Distribution	Original	Duplicate
Population	271		25.0%	0.02	0.02
Minimum (g/t)	0.01	0.01	50.0%	0.03	0.03
Maximum (g/t)	5.82	5.95	75.0%	0.06	0.05
Mean (g/t)	0.42	0.42	80.0%	0.11	0.12
Std Dev	0.86	0.85	90.0%	0.19	0.19
CV	2.03	2.04	97.5%	0.26	0.26
Correlation	0.998		99.9%	0.38	0.35

The correlation plot presented in Figure 11-19 shows a good correlation with an evenly distributed grade in both original and duplicates. Most of the Duplicates grades are noted to be well within 10% threshold.

From scatter plots analysis (Figure 11-20) a good correlation is observed with most samples falling within a 10% acceptable range. The data presents a correlation coefficient of 0.998 which is deemed acceptable.

Precision (Figure 11-21) shows that 87% of the coarse duplicates analysed at SGS Doko returned paired results within a 20% difference of each other and 70% of the Trench field duplicates within 10% difference. Usually 80% of data within 20% difference with a threshold of 64% of pairs within 10% difference is deemed acceptable. The overall performance is in line with the expected threshold.

18^th September 2018 Page 107

Kibali Gold Mine NI 43-101 Technical Report

Figure 11-19 Ascending Grade Correlation Plot for SGS Doko Trench Field Duplicates Between 0 g/t Au to 10 g/t Au

Figure 11-20 Normal Scatter Plot of SGS Doko Trench Field Duplicates £ 10 g/t Au Tail Cut

18^th September 2018 Page 108

Kibali Gold Mine NI 43-101 Technical Report

Figure 11-21 Precision Plot of Trench Field Duplicates vs Original Sample at SGS Doko

Umpire Assays

Pulp duplicate samples are submitted to an external independent laboratory for umpire analysis. The ALS OMAC (OMAC) laboratory based in Ireland is used as the independent umpire laboratory and samples are submitted biannually from Kibali. CRM samples are submitted along with umpire samples to check for bias at the umpire laboratory.

Pulp samples submitted to OMAC were a second test of where there could be possible bias in assays received from SGS Doko. A total of 1,625 pulp duplicates plus check samples were submitted from different grade ranges which show some bias between SGS Doko and OMAC (Table 11-14).

Figure 11-22 illustrates that OMAC is consistently higher from 2.5 g/t Au to 6 g/t Au and above 7.5 g/t Au compared to SGS Doko but within 10% tolerance, which is considered as acceptable, however, Kibali plan to undertake further umpire analysis at a third laboratory to verify the results. In the QP’s opinion, this is not considered as having a material impact on the relative accuracy of the Mineral Resource.

18^th September 2018 Page 109

Kibali Gold Mine NI 43-101 Technical Report

Table 11-14 Summary of Pulp Duplicates Analysed at OMAC

Statistics	Original	Duplicate	Distribution	Original	Duplicate
Population	1,625		25.0%	0.03	0.01
Minimum (g/t)	0.01	0.01	50.0%	0.04	0.02
Maximum (g/t)	11.50	14.80	75.0%	0.07	0.05
Mean (g/t)	0.95	1.01	80.0%	0.14	0.13
Std Dev	1.85	2.04	90.0%	0.27	0.27
CV	1.95	2.02	97.5%	0.58	0.62
Correlation	0.959		99.9%	0.90	0.93

Figure 11-22 Normal QQ Plot of Pulp Duplicates SGS Doko vs. OMAC at 10 g/t Au Tails Cut

11.5 Security

Samples are under security observation from collection at rig, to processing at the site core yard, to delivery at the laboratory.

RC samples are weighed and documented on the rig.

Samples, including duplicates, were delivered from the drill rig to a secure storage area within the fenced Kibali core facility. Then blanks and certified reference materials were inserted.

Labelled samples are placed into large bags and sealed. The large bags are placed in a crate which are transported to the warehouse and trucked to the relevant laboratory by Kibali Goldmines personnel.

Chain of custody procedures consisted of filling out sample submittal forms that are sent to the laboratory with sample shipments to make certain that all samples were received by the

18^th September 2018 Page 110

Kibali Gold Mine NI 43-101 Technical Report

laboratory. Sample security has relied upon the fact that the samples are always attended or locked in appropriate sample storage areas prior to dispatch to the sample preparation facility.

Analytical results from all laboratories are emailed to a Project email group and are later imported into the database by the Database Administrator. A paper certificate is mailed at a later date.

Pulp samples are stored under conditions that are kept clean and dry to avoid contamination.

11.6 Independent Audits

Optiro completed an audit of the Mineral Resource and Ore Reserve processes in 2017, which found no essential issues to be fixed. A number of recommended issues were identified but have all been implemented or are in the process of being implemented. These include:

☐ A new Gilson splitter has been installed at all RC drilling sites for better capacity and accuracy in collecting and splitting samples.

☐ Geologist to select location of blank insertion based on the mineralisation, compared to a blank every 20 samples, which results in missing the mineralisation.

☐ The DD Core duplicate was collected by laboratory initially, so the sample was known to the lab, however as per recommendations, the Kibali technician is going to laboratory to collect coarse reject and re-bag it as duplicate.

☐ The introduction of data loggers, specifically for capturing sampling information and for auto-generation of sample numbers.

☐ The relatively poor correlation between pulps from SGS and those from the check laboratory, ALS (OMAC) should be investigated. Kibali will be using a second check or umpire laboratory for the next series of samples.

☐ The RC subsampling practice needed to be revised to ensure consistent sample weights within the tolerance (3.0 kg to 3.5 kg) required by the SGS Doko assay laboratory. A new set of Gilson splitters were procured and installed onsite, thereby ensuring an unbiased and consistent subsample.

☐ In line with the above, once RC samples of the right mass are being delivered to the laboratory they should all be processed through the Boyd crusher and associated rotary sample divider. The laboratory should not need to split the samples using a riffle splitter as is current practice.

☐ Blank samples (granite fragments) should be selectively inserted within the high-grade intersections as logged, whether for diamond core or RC chips. This will ensure maximum effectiveness for the blank insertion process.

11.7 Discussion

Assays from both SGS Doko and Mwanza laboratories are generally within global acceptable accuracy, with low, medium and high-grade ranges showing acceptable level of match and within 10% difference. All blanks analysed by SGS Doko and Mwanza are well within the acceptable limit. Few batches were re-analysed and re-assayed values were well within two standard deviations.

18^th September 2018 Page 111

Kibali Gold Mine NI 43-101 Technical Report

Field duplicates assayed by SGS Mwanza show very good correlation and repeatability, even though from the scatter plots there are certain grade ranges that show some level of bias, this can be expected, and hence does not have very high impact on the overall reported grades. Field duplicates from SGS Doko, at times show poor correlation and repeatability, there are certain grade ranges that show some level of biasness. The sampling and splitting method have been revised and improved from what we learn on quarterly basis thus overall performance improved. Sampling and splitting method are continuously monitored to improve the overall performance.

A low-level bias in few duplicate samples at very low-grades (<0.1 g/t Au) has been identified and should have no material impact on the resource and grade control estimate. Also noted from the field duplicate is the fairly high nugget effect (+30%) that has been associated with grade distribution with the deposit where grade control drilling has been completed. This is currently matching what is physically observed in most drill areas of low-grade intercalations associated with specific rock units (schist).

CRMs have performed well, and the overall performance is acceptable. All low, medium, and high-grade samples showed acceptable levels of performance and therefore acceptable levels of accuracy. There were few failed standards within the period under review with low-grade ranges which have been investigated and removed from the sample stream not used anymore.

Blanks have performed very well within the period under review and thus the material used for blanks is deemed acceptable

A review of the QA/QC data within the period has indicated that 98.4% and 99.9% of the Standards and Blanks respectively inserted into the sample streams return values within the 10% acceptable limits.

Field duplicates pairs compare well. Revised approach has been further enforced and expected to get even better. A new Gilson sample splitter was recommended by Optiro and subsequently implemented by Kibali at all RC drilling sites. This splitter is of larger capacity and more reliable in obtaining unbiased 50:50 samples.

Overall the QA/QC results returned are acceptable, however they can be improved with implementation of stringent laboratory protocols and procedures, such as a full implementation of a LIMS sample submission and results reporting system to complement the existing LIMS tracking system.

Independent laboratory check samples are sent to OMAC laboratory in Ireland on a bi-yearly basis. The last three batches of pulps duplicate results received, do not show very good correlation between SGS laboratory at Doko and OMAC at Ireland. OMAC pulp duplicates and laboratory repeats all show consistent drift above 10% from the original assays, CRM on the other hand have mostly been under reported compared the expected values. This suggests that OMAC laboratory are under reporting the results. Consequently; in Q1 2018, 1,200 pulp duplicate samples will be sent to ALS Johannesburg as third laboratory check.

18^th September 2018 Page 112

Kibali Gold Mine NI 43-101 Technical Report

Overall the QA/QC results returned are acceptable, however they can be improved with implementation of stringent laboratory protocols and procedures, such as a full implementation of a LIMS sample submission and results reporting system to complement the existing LIMS tracking system.

In the QP’s opinion, the sample preparation, analysis, and security procedures at the Kibali are appropriate for use in the estimation of Mineral Resources.

18^th September 2018 Page 113

Kibali Gold Mine NI 43-101 Technical Report

12 Data Verification

12.1 Historical Drill Hole Data Verification

Cube Consulting Pty Ltd (2009) report that they validated the historical drill hole data and provided the following comments.

☐ Very little information was available regarding drilling by OKIMO, Barrick or AngloGold Ashanti.

☐ Diamond Core (DD) and Reverse Circulation (RC) drilling was undertaken on behalf of Moto by GeoSearch Limited (GeoSearch), an experienced drilling contract company sourced from South Africa. At the time of the study in 2009 there were up to four DC rigs and one RC rig operating at the Moto Gold Project.

12.2 Kibali Drill Hole Data Verification

All forms of Project data are stored secured in industry standard Maxwell Geoservices (Maxwell) DataShed SQL database. Data must pass validation through constraints, library tables, triggers, and stored procedures prior to importing. Failed data is either rejected or stored in buffer tables awaiting correction. A full-time database administrator employed at site manages the database.

Daily and weekly backups are made and stored on site. Copies of monthly back-ups are sent to Entebbe and quarterly backups are sent to Johannesburg.

A custom MS Access front end application has been designed for data entry, reporting, and viewing via Open Database Connectivity (ODBC), which utilises the data validation procedures from the SQL database. All other geological and mining software databases on site use ODBC link to retrieve information from DataShed SQL database.

Assay data is imported directly from assay certificates from the laboratory and validated. Only fully trained and authorised network users can upload laboratory data. Assay data is stored in a normalised format and multiple assays are stored for each sample. Ranking of different assay formats is performed automatically so that one assay result is displayed in the tblVWDHAssays table. Any change to the rankings (held within tblSYSAssMethod) must be approved by the onsite Database Manager.

Twin holes are used to access the accuracy of some intersections that require further details.

In the QP’s opinion, the Kibali Mineral Resource databases are considered to be appropriate to be used for the estimation of Mineral Resources and Ore Reserves.

18^th September 2018 Page 114

Kibali Gold Mine NI 43-101 Technical Report

12.3 Independent Audit

An independent external database audit was completed by Maxwell in March 2016. Maxwell identified that the majority of resource data within the SQL database was in good order and only minor data issues were identified.

Issues outlined by Maxwell included, assays that were unranked, 339 QC records with unmatched drill records, and mismatched drill dates between tables. All data that was flagged as having minor issues was quarantined and corrected before being released from the quarantine table by Kibali.

Continued training and mentoring are ongoing for the database administrators as was recommended by Maxwell.

Following the external audit of the Kibali database, a JORC (2012) Code compliance certificate was issued by Maxwell.

18^th September 2018 Page 115

Kibali Gold Mine NI 43-101 Technical Report

13 Mineral Processing and Metallurgical Testing

13.1 Summary

There have been a number of testwork programs completed on Kibali ore. Some testwork was performed to help validate the results of historic studies. Testwork programs for some satellite deposits were completed subsequent to initial plant commissioning or other targeted characterisation. A summary of the testwork to date can be found in Table 13-1.

Table 13-1 Summary of Testwork

Name of Program	Laboratory	Report ID or Number	Publication Date
Metallurgical Testwork Including Risk Reduction and Variability Tests	AMTEC/OMC	A12949 A TO D	2011
Bankable Feasibility Study	Amtec/Oway Minerals Consultants/Senet Engineering (SA)	Senet        Kibali Goldmines Feasibility Report	2010
Feasibility Study	AMTEC (Now ALS)/Lypocodium Engineering	1329/16.15/1329- STY-002/S5-B	2007
Prefeasibility Study	AMTEC (Now ALS)/Lypocodium Engineering	1329/16.15/1329- STY-001/S5-B	2006
Satellite Pits and Additional Work
Mengu Hill
Deportment of Gold in Mengu Hill feed and flotation products	AMTEL	Amtel        Report 12/55	2013
Mengu Hill Testwork Summary (Appendices available with all details of sample selection and compositing strategies)	AMTEC/OMC	Report No.   8888 Rev 1	2012
Pakaka
Metallurgical Performance of the Pakaka Feed Blends in the CIL – Review Relative to Feasibility and Geomet Arsenic domains	Kibali Goldmines Internal Review and Geomet Report		2017
Laboratory flotation testwork on Pakaka gold samples (also includes in APP reports work on mineralogy)	Outotec Research Finland	15142-ORC-T	2016
Gold deportment analysis of Pakaka major ore types	AMTEL	Amtel        Report 14/14	2014
Gorumbwa
Metallurgical Testwork conducted upon samples from the Gorumbwa Project for Kibali	ALS Metallurgy (Formerly AMTEC)	Report        No. A16184	2016
Gorumbwa Feasibility Study – Metallurgical Testwork Report	Kibali Goldmines Internal Review and Summary of all tests conducted - T. Mahlangu	Internal Report	2014
Gold Deportment in Gorumbwa ores by CN leach	AMTEL	Amtel        Report 14/42	2014
Sessenge
Processing of three samples from the Kibali - Sessenge Pit according to the current Kibali flowsheet	Maelgwyn Mineral Services Africa	REP 18-008	2018
Kibali Met Laboratory Sessenge Geomet Work_2018	Kibali Geomet Internal Testwork and Review Report	Internal         Report	2018
Deportment of gold in Kibali Sessenge ores	AMTEL	Amatel Report 16/38	2016
Pamao
Pamao Gravity Testwork	Peacocke & Simpson	PS394A to F	2017
Pamao BRT and Arsenic Distribution	Kibali Geomet Internal Review Report	Internal Report	2017

18^th September 2018 Page 116

Kibali Gold Mine NI 43-101 Technical Report

Name of Program	Laboratory	Report ID or Number	Publication Date
Metallurgical Testwork – Pamao_2017	Kibali Internal Pamao Metallurgical Review - T. Mahlangu	Internal Report	2017

Summary of Results

The extensive metallurgical testwork campaigns demonstrate two distinct behavioural patterns where some ore sources, in particularly the oxides, with some fresh rock sulphide sources also, exhibit free-milling characteristics suitable for gold extraction by a conventional CIL metallurgical process. Other ore sources exhibit a degree of refractoriness, albeit never extreme, where straight cyanidation returns gold dissolutions in the region of 70%, considered to be too low for optimal plant operation. This refractoriness is invariably the presence of occluded gold particles within sulphide minerals. It has been determined that a finer grind will expose a portion of this additional gold for leaching, thus enhancing the recovery such that it exceeds 80%. In addition many of the Kibali ore sources, exhibit a preg-robbing tendency, which points to the need for rapid carbon adsorption. Thus, the Kibali plant design was to cater for these observations through two distinct processing circuits:

☐ Free-milling ore sources – conventional CIL circuit.

☐ Refractory ore sources – flotation circuit with ultra-fine-grinding (UFG) and dedicated intensive leaching of the concentrate generated. Float tails leaching is optional and dependent on both of:

o The current holding capacity of leached tails i.e. the CTSF or cyanide tailings storage facility.

o High flotation recoveries, often render float tails leaching as uneconomic.

o Use of flotation tails in backfilling will require a detox step thus potentially rendering the whole float tail leach step uneconomic.

More detailed descriptions of the discreet metallurgical testwork campaigns follow.

Open pit extraction variability was low between the two processes at 1.62% but with a lower average extraction of 83.7% for the KCD.

The LOM average gold extractions are 89% excluding the leach tails with minimum and maximum recoveries of 78.4% and 96.4% respectively.

Samples with high extraction variances have been isolated and further analysis has revealed the following:

☐ The exclusive use of extraction variability to decide on whether to include or exclude float tails leach is inadequate and can be misleading.

☐ Poor dissolution of concentrate and leach of float tails suggests that there is a benefit in leaching float tails when in fact gold losses occur in the float concentrate leach residue.

☐ Poor recoveries to flotation concentrate results in high float tails residues (>1 g/t) even when high dissolution levels are achieved in the concentrate leaching.

18^th September 2018 Page 117

Kibali Gold Mine NI 43-101 Technical Report

☐ The average float tail grade for the samples that presented high variability was 0.395 g/t Au with a minimum of 0.04 g/t Au and maximum of 1.32 g/t Au.

☐ A strong impact on feed or head grade which translates into high flotation concentrate grades leading to high residue tails, residence time and reagents availability issues.

☐ Subsequent plant operational data has demonstrated consistent sub 0.1 g/t flotation residues and this has completely eliminated any need to leach flotation tails from fresh sulphide ore sources.

The resultant strategy is to:

☐ Maximise gold recovery into the flotation concentrate – less through increased mass pull, due to the capacity limitations imposed by the downstream concentrate treatment processes in particular UFG, and rather by reagent suite optimisation including optimal and steady flotation operation.

☐ Maximise gold dissolution from the concentrate – mineralogical effects might have an effect, but regular diagnostic leach tests will help keep track and identify where the problems come from.

☐ Additional residence time for concentrate can be provided by the CIL – pumpcell product is provided for the benefit of further gold dissolution in the larger tanks.

Gold Recovery

The concise explanation of the testwork programme, results and interpretation of these results are detailed in the Amtec Reports (A12949 Parts (A – D) Full Report) covering the:

☐ Metallurgical samples characterisation in terms of grade, mineralogy, and physical characteristics of crushing and grinding parameters.

☐ Experimental procedures, collation, and analysis of leach test results.

☐ Extraction variability tests as well as comminution variability tests.

There is extensive laboratory testwork data available completed under the Moto Feasibility studies which has been used to complement the extraction variability tests on the main composite samples.

13.2 Testwork Strategy & Sample Selection: Extraction

The physical and extraction sample selection and testwork logic was developed by Lycopodium and used in the feasibility and optimised feasibility for Moto Gold ores (Table 13-2). Extraction results that are presented in the figures below include the OFS results and extraction variability tests conducted in 2010.

A total of 136 drill hole composite samples, composited at 10 m to 12 m interval, were subjected to direct cyanidation. The test procedure involved milling the samples to 80% passing 75 microns, bottle roll leaching in the presence of oxygen at 40% solids, pH 10.5 and 0.2%w/v nominal strength of cyanide for 24hrs. Note that the Master Composite and extraction variability were selected for detailed metallurgical investigation based on the geological description of the oxidation state and not the metallurgical behaviour of the hole composite samples.

18^th September 2018 Page 118

Kibali Gold Mine NI 43-101 Technical Report

The results, as depicted in Figure 13-1 indicated significant spatial changes in the cyanidation response of the deposit. The scattered nature of the results indicated that certain samples logged as primary material responded very positively to direct cyanidation.

Figure 13-1 Initial Hole Composite Dissolutions (Blue Markers = Fresh, Green Markers = Transition, Red Markers = Oxide)

Besides the extraction variability samples, metallurgical testwork was conducted on the risk reduction samples, for both oxide and sulphide/fresh samples. The results from these tests are also included in the figures that follow. The data represented in Figure 13-2 gives the extraction variability for the primary process of gravity float – float concentrate leach with the exclusion of flotation tails. EV2009 and EV2010 are the extraction variabilities for the (2009) Feasibility study and the 2010 testwork on the fresh material. Also, OFS_UG 1 to 7 represents the underground samples for the Optimised Feasibility Studies.

The average extraction of all fresh samples, that is, open pit and underground excluding the leaching of tails, is 88.1%. Also included in the plots are the underground feasibility recovery (89.8%) and open pit feasibility recovery (86.1%). Except for the OFS_UG samples, the extraction data is plotted as a function of the Diamond Drill Holes.

18^th September 2018 Page 119

Kibali Gold Mine NI 43-101 Technical Report

Table 13-2 Physical and Extraction Sample Selection and Testwork Logic (Moto FS_2007- 1377\16.14\1377-STY-002\S5)

Select Drill Hole Metre Interval Intercepts from Site
Select Comminution Testwork Samples and Comminution Variability Samples
Select samples for mineralogical thin section investigation
Conduct JK Drop Weight tests, Apparent SG, Abrasion index, BWi, RWi, SMC tests and Levin open circuit grindability testwork on selected samples.
Crush remainder (P100 2 mm), Mix, Split, Assay and Leach a sub-sample of each 10 m Interval for each hole to determine direct cyanidation characteristics of individual hole composites
Select Master Composite samples for Primary and Oxide material individually
Select Variability samples, both spatially and by rock type
Oxide Master Composite	Primary Master Composite
Head Assays	Head Assays
Mineralogical Investigation	Mineralogical Investigation
Grind optimisation and leach tests	Grind optimisation and flotation tests
Gravity gold recovery, including intensive cyanidation of gravity concentrates at “as received” and ultra-fine-grind.	Gravity gold recovery, including intensive cyanidation of gravity concentrates at “as received” and ultra-fine-grind.
Leach optimisation, including reagents, oxygen vs. air sparging, diagnostic analysis and retention time	Direct cyanidation tests, including reagents, oxygen vs. air sparging, diagnostic analysis and retention time
Flotation testwork	Flotation reagent optimisation testwork, including flotation tests in site water
Oxygen Uptake Rate determination	Bulk gravity separation and pilot flotation
Viscosity measurements at varying pulp densities	Flotation Tail
Flocculation and thickener testwork	Head assays
Sequential Triple Contact CIP (Carbon-in-pulp) testwork and Equilibrium Carbon Loading testwork	Leach tests
Geochem analysis on leach tail	Viscosity measurements at varying pulp density
Cyanide Detoxification testwork	Thickener and flocculation testwork
	Geochem analysis
	Flotation Concentrate
	Head assays, true SG determination, mineralogical examination
	Ultra-fine-grind testwork and leach optimisation, including reagents, oxygen vs. air sparging, and retention time
	Indicative oxidation testwork: - Pressure Oxidation, - Roast Calcination, - Bio-oxidation, - Albion Process
	Oxygen Uptake Rate determination
	Viscosity measurements at varying pulp densities
	Flocculation and thickener testwork
	Sequential Triple Contact CIP testwork and Equilibrium Carbon Loading testwork
	Geochem analysis on leach tail
	Cyanide Detoxification testwork
Upon completion of the extraction testwork, the Process Route is defined for Oxide and Primary Material
Subject Primary variability samples to optimal recovery conditions as determined for the Primary Master Composite material
Subject Oxide variability samples to optimal recovery conditions as determined for the Oxide Master Composite material
Subject Transition variability samples to optimal recovery conditions as determined for the Primary Master Composite material

18^th September 2018 Page 120

Kibali Gold Mine NI 43-101 Technical Report

Figure 13-2 Primary Extraction Excluding the Leaching of Flotation Tails

The OFS_UG variability samples were composited according to Table 13-3 the results of which are not characteristic of each drill hole, but of the composite.

18^th September 2018 Page 121

Kibali Gold Mine NI 43-101 Technical Report

Table 13-3 Extraction Comparison - Underground Variability

Sample Number	Hole Number	Intercept		Oxidation State	Source	Extraction (Oxide Process Route) % of Total Au	Extraction (Primary Process Route) % of Total Au	Extraction (Primary Process Route plus Flotation Tail Leach) % of Total Au
		From   (m)	To (m)
1	DDD22   8	507	52   8	Fresh	Lode 910	75.6	86.3	89.8
	DDD08   0	537	58   8
	DDD22   7	546	55   4
	DDD21   3	482	49   6
	DDD12   8	471	47   5
2	DDD20   7	474	48   8	Fresh	Lode 910	76.5	90.8	92.6
	DDD21   2	468	51   8
	DDD29   3	472	48   0
	DDD29   7	476	48   8
	DDD30   6	530	54   2
3	DDD29   7	474	49   4	Fresh	Lode 910	72.3	84.2	86.1
	DDD07   3	446	45   2
	DDD21   9	488	51   2
	DDD13   0	482	51   6
4	DDD22   5	544	54   6	Fresh	Lode 910	76.5	86.4	91.1
	DDD03   1	646	65   0
	DDD17   5	476	48   8
	DDD21   1	480	51   2
5	DDD27   1	472	49   2	Fresh	Lode 910	71.2	87.7	90.4
	DDD22   7	592	61   2
	DDD26   9	663	67   5
	DDD20   7	488	50   2
	DDD10   3	308	33   0
6	DDD12   9	542	57   0	Fresh	Lode 910	78.9	92.1	94.5
	DDD06   9	470	48   8
	DDD24   0	484	49   8
	DDD20   6	481	51   9
7	DDD07   0	480	48   8	Fresh	Lode 910	60.9	91.1	93.3
	DDD08   5	438	45   6
	DDD20   5	504	52   6
	DDD21   4	508	54   2
Master Composite				Fresh	Lode 910 & 920	76 .9	91.3	93

18^th September 2018 Page 122

Kibali Gold Mine NI 43-101 Technical Report

Extraction variability data with the leaching of flotation tails is shown in Figure 23. The benefit and non-benefit of leaching flotation tails can be analysed in terms of the variance in recovery between the two process routes as well as other factors that include, gold recovery into the flotation concentrate and concentrate residue values. The variance for this data is shown in Figure 13-3. There is significant variance for the data, as presented, and this data is a combination of both open pit and underground samples.

Figure 13-3 Primary Extraction Variability Including the Leaching of the Flotation Tails.

Although these samples are considered fresh, it is apparent that their leaching response differs. What is not clear is why some respond positively to the leaching of flotation tails and meaning that there are non-floating minerals such as silicate mineral that contain gold or alternatively there is depression of passivated minerals during flotation. Samples in Table 13-4 were isolated for further analysis to define the reasons causing the variances, also depicted in Figure 13-4.

Table 13-4 Isolated Samples for Further Analysis

No.	Drill Hole	From	To	Lithology	Lode	Variance
1	DDD011	249	263	acs(vag)	j-3000	12.77
2	DDD224	101	114	vag/cs	i-3000	7.72
3	DDD005	450	464	vag/tuff	ug-9000	9.45
4	DDD290	294	308	vag/cs	ug-9000	8.44
5	DDD211	546	560	vag/xtaltuff	ug-9000	7.45
6	DDD127	496	510	is/vag	uga-9000	6.91
7	DDD084	184	198	is/sbx	ia-5000	6.71
8	DDD195	150	164	vag	n-5000	4.88
9	DDD162	40	54	vag	m-5000	4.72
10	OFS_UG 4				Lode 920	4.7
11	2010 MC					4.61

18^th September 2018 Page 123

Kibali Gold Mine NI 43-101 Technical Report

Figure 13-4 Extraction as a Function of Diamond Drill Holes

Figure 13-5 Analysis of the Drill Hole Samples Exhibiting Large Variances

The results (Figure 13-5) represent the variation of gold recovery into the flotation concentrate (top left), initial sample head grade (top right), concentrate grade (bottom left) and float tail and concentrate leach residue (bottom right) for the drill holes with relatively high variances between the two processing routes.

18^th September 2018 Page 124

Kibali Gold Mine NI 43-101 Technical Report

Gold losses hence variances between the two process routes can be defined in terms of (i) losses to the flotation tails, that is, poor flotation recovery or (ii) losses to the concentrate leach residue.

In the first case, the presence of non-floating or slow floating mineralised material in the ore leads to the observed gold loses and is the reason why the leaching flotation of tails provides an improvement in gold recovery. This requires an optimised flotation circuit. From the available data, low-grade ore samples such as DDD224 and DDD005, present poor flotation recoveries which transcends into poor overall recovery on the exclusion of the leaching of the flotation tails The split into the float concentrate and tails is around 7:93 (7% mass pull), leaching of the low-grade flotation tails will compensate for any losses into the concentrate residue. The fundamental point lies with the maximisation of flotation recovery and subsequent dissolution of gold from the flotation concentrate.

The gold losses into the concentrate leach residue is either a function of inadequate reagents or residence time on treating high-grade flotation concentrates (DDD162) or mineralogical effects related to gold occlusion in the form of finely disseminated particles such that even ultrafine grinding does not liberate it.

The results in Figure 13-6 were extracted from the OMC report and include the oxide and transition material leach (red plots) results. These results are included in this analysis for comparison purposes.

Figure 13-6 Gold Extractions Obtained for Various Extraction Variability Tests and Master Composite Samples (OMC report).

18^th September 2018 Page 125

Kibali Gold Mine NI 43-101 Technical Report

Extraction Results: Oxides, Transition and Other Ores

The average recovery on this data is 89.9% with a minimum value of 78% and maximum of 99.12% (Figure 13-7). The direct leach tests results from the grade control samples are shown in Figure 13-8. The results have an average recovery of 90% with a range of 82% to 96%.

Figure 13-7 Plots of Extraction Using the Primary Process for the Oxide Materials – KCD

Figure 13-8 Direct Cyanidation of Grade Control Samples

18^th September 2018 Page 126

Kibali Gold Mine NI 43-101 Technical Report

It is clear from the two sets of data that the strategy adapted for the treatment of the oxide/transition material of the KCD is sufficient to minimise any gold losses. The benefit of treating the transition material through the oxide route with the flash flotation component ensures that both sulphides and non-floating materials are treated in the UFG – leach and CIL respectively.

The leach results for the gravity – direct cyanidation tests on the grade control samples is detailed in the Table 13-5.

Table 13-5 Direct Cyanidation Results

Sample No.	Assay Head (g/t Au)	Calc Head (g/t Au)	Solids Tail Value (g/t Au)	Gravity Recovery (%)	Dissolution (%)	Total Extraction (%)	Lime Cons (kg/t)	NaCN Cons (kg/t)
DCRC0049 4.0-1 4.0 m	10.3/10.7	10.2	1.37	27.12	59.49	86.61	0.56	0.51
DCRC00 37 22.0-32.0 m	18.6/18.2	20.3	0.91	17.22	78.3	95.52	0.68	0.55
DCRC0047 14.0-24.0 m	9.08/9.28	9.75	0.96	17.69	72.46	90.15	0.59	0.70
DCRC00 50 4.0-14.0 m	12/11.2	12	1.19	33.80	56.26	90.06	0.55	0.65
DCRC0008 26.0-36.0 m	5.27/4.81	5.36	0.23	25.21	70.5	95.71	0.46	1.46
DCRC0007 46.0-58.0 m	2.79/2.26	2.76	0.22	13.72	78.31	92.03	1.79	0.82
DCRC0047 4.0-14.0 m	3.56/3.7	3.6	0.51	15.42	70.42	85.84	0.83	0.76
DCRC0040 16.0-26.0 m	1.63/1.83	1.75	0.21	8.46	79.56	88.02	0.53	0.87
DCRC0046 4.0-14.0 m	1.8/1.76	1.75	0.31	10.11	72.14	82.25	1.23	0.67
DCRC000S 8.0-1 8.0 m	0.8/0.72	0.84	0.03	16.07	80.36	96.43	0.75	0.38
DCRC0013 68.0-78.0 m	0.56/0.44	0.58	0.05	3.45	87.93	91.38	0.87	1.26

13.3 Open Pit Operations

With Pakaka pushback 1 scheduled for completion during Q1 2018, the processing of Sessenge as an additional satellite pit commences in 2018. A similar Geometallurgical work programme to that of Pakaka, has been conducted on Sessenge. Three geometallurgical significant lodes have been identified, namely the two high-grade shoots, 9102 and 9103 and the low-grade envelope, named as 9002. The fresh ore within the high-grade shoots have been characterised with poor recoveries, additional work is in progress to define the arsenic model. The primary objective is to generate simplified models for:

☐ Arsenic – Grade – Recovery based on bottle roll tests

☐ Domains based on defined lodes or high-grade chutes and main envelopes

The oxide and transition ore recoveries have returned an upside from feasibility testwork. Because of the drop in the overall fresh recoveries compared to the feasibility tests and outstanding arsenic assays to complete the arsenic model, additional testwork is ongoing and

18^th September 2018 Page 127

Kibali Gold Mine NI 43-101 Technical Report

will be completed by June 2018 to clearly define all the opportunities and gaps that could improve the fresh ore recovery. Figure 13-9 shows the preliminary Sessenge Geomet Model, defining the various domains identified to date. The main objectives are to clearly define and confirm the grade – arsenic – recovery, domains that will feed into an optimised strategy.

The Sessenge reserves are based on the feasibility metallurgical recoveries awaiting the finalisation of the arsenic model. This will be updated and incorporated in the mine plan before mining commences in late Quarter 1 2018.

Figure 13-9 2016 Sessenge Geomet Model

Additional geo-metallurgical testwork on KCD Pushback 3, confirmed the oxide recoveries and re-defined the reagents consumptions and blending strategies with the current suite of free milling ores from Pakaka and Kombokolo (Table 13-6). Overall, the current geo-metallurgical work is focused on closing gaps and continuous improvement to enhance efficiencies in terms of recoveries and reagents consumption. These have been built into the budget models to ensure that operating costs and efficiencies are managed live to reflect the current ore feed blends.

18^th September 2018 Page 128

Kibali Gold Mine NI 43-101 Technical Report

Table 13-6 Metallurgical Recoveries Used for Deposits in Pit Optimisation

Ore Source	Recovery
	Oxide (%)	Transitional (%)	Fresh (%)
KCD	90.1	90.1	86.1
Sessenge	90.3	75.9	79.1
Pamao	90.9	85.0	83.5
Kombokolo	85.0	85.0	85.0
Gorumbwa	90.0	90.0	90.0

Variability of Results and Confidence Levels

From the data available the metallurgical sampling and extraction testwork covered the ore bodies of KCD, Kombokolo, Mengu Hill, Pakaka, Pamao, and Sessenge. While all the ore samples have been tested, the selection of the process routes and subsequent plant design has been based on the results from the KCD which consists of 70% of the feasibility feed to the plant. The most significant increase in tonnage is likely to come from the KCD deposit. The sampling strategy and classification of samples in the KCD area has followed in principle the process depicted in Figure 13-10.

18^th September 2018 Page 129

Kibali Gold Mine NI 43-101 Technical Report

Figure 13-10 Sampling Strategy and Classification of Samples in the KCD

18^th September 2018 Page 130

Kibali Gold Mine NI 43-101 Technical Report

The KCD Metallurgical Analysis and Extraction Data can be found in Table 13-7.

The selection was similarly performed for all ore bodies, either open pit or underground and based on redox state as oxides, transition, or fresh.

The process routes selection strategy ensures that fast floating material and gravity recoverable gold are recovered upfront with minimal size reduction requirements. Provision has been made for this in the form of, included in both process streams, flash flotation and gravity concentration units. The benefits and possible shortcomings of this arrangement have been briefly discussed in the section covering the strategy around handling oxide/transition material classification.

Table 13-7 KCD Fresh Open Pit Fresh Samples - Lode 5000

Hole ID	Sample   ID	From   (m)	To   (m)	Weathering	UG/OP	Lode	Overall Extraction (excl float tail leach) %	Overall Extraction (incl float tail leach)  %
DDD072	EV2009	120	130	Fresh	OP	5000	89.8	92.8
DDD257	EV2009	140	170	Fresh	OP	5000	81.48	83.64
DDD165	EV2009	86	96	Fresh	OP	5000	84.13	85.51
DDD165	EV2009	116	126	Fresh	OP	5000	81.76	82.91
DDD160	EV2009	90	100	Fresh	OP	5000	78.38	79.44
DDD195	EV2010	150	164	Fresh	OP	5000	90.67	95.55
DDD162	EV2010	40	54	Fresh	OP	5000	92.01	96.73
DDD166	EV2010	96	110	Fresh	OP	5000	92.4	93.97
DDD164	Master   Comp	92	110	Fresh	OP	5000
DDD455	Master   Comp	113	148	Fresh	OP	5000
DDD424	Master   Comp	54	89	Trans	OP	5000
Master Comp Final							87.3	92.35

18^th September 2018 Page 131

Kibali Gold Mine NI 43-101 Technical Report

13.4 Metallurgical Recoveries

The sample selection for the ore bodies and metallurgical recoveries expected at Kibali and used in the financial model can be found in Table 13-8.

The samples have been selected by site geologists and metallurgists and, in the opinion of the QPs, are representative of the ore bodies across the Permits.

Table 13-8 Summary of Average Recovery for All the Samples

Ore Source	Weathering	Average Recovery Primary Only (%)	Average Recovery Primary + Tail Leach (%)	Average Recovery Oxide Process (%)	Feasibility or Financial Model Recovery (%)
KCD	Fresh_OP	86.4	89.2		86.1
	Fresh_UG	89.0	93.4		89.8
	Transition	66.6	91.3		90.1
	Oxide			89.1	85.8
Sessenge	Fresh	72.7	81.2		79.1
	Transition		80.3		75.9
	Oxide			90.4	90.3
Pakaka	Fresh	78.1	82.3		80.2
	Transition				81.3
	Oxide			96.9	88.7
Mengu Hill	Fresh	69.2	72.2		70.1
	Transition	84.4	89.9		89.3
	Oxide			92.6	89.3
Kombokolo	Fresh	70.3	75.2		73.1
	Transition	78.9	95.3		95.9
	Oxide			96.4	95.6
Pamao	Fresh	74.5	85.5		83.5
	Transition				85.0
	Oxide			95.8	90.9

13.5 Deleterious Elements

Kibali needs to consider the remediation of cyanide species as well as arsenic.

Cyanide

Kibali chooses to abide by the guidelines of the International Cyanide Code. AngloGold Ashanti is a formal signatory, and Kibali follows the requisite cyanide protocols. The cyanide tailings storage facilities (CTSFs), of which there are two, have both been lined with an HDPE liner. Protocols call for limited threshold discharges to the CTSF and cyanide discharge concentrations are controlled through use of an on-line cyanide analyser and controller. The presence of two CTSFs allows management of the cyanide containing liquor streams and moreover, most of the water is recycled to the plant area where there also exists an additional cyanide detoxification pond facility.

Arsenic

The main deleterious element in the Kibali ore sources is considered to be arsenic. Certain isolated ore types exhibit higher levels of arsenic (in Pakaka and Sessenge) which can result in dissolution

18^th September 2018 Page 132

Kibali Gold Mine NI 43-101 Technical Report

during the recovery process. The impact of arsenic is in the leach of flotation concentrate in the intensive oxygenation/cyanidation circuit.

Mitigation can occur for either of the cyanide containing streams or non-cyanide containing streams i.e. flotation tails, which reports to a dedicated but unlined flotation storage facility (FTSF). Arsenic remediation can occur through oxidation of ferrous sulphate and arsenic species to the valency state (V). Alternatively, ferric chloride may be used directly, though is associated with corrosion issues. Both methods result in the formation of a stable ferric arsenate precipitate. The primary mitigation method utilised at Kibali is the application of a blending strategy where high arsenic content ores are intentionally blended with ores with low content, thereby restricting the arsenic solution tenors within the circuit.

Arsenic content in excess of 2,000 ppm has a negative effect on gold dissolution where dissolution values as low as 70% are attained when arsenic content increases to values as high as 9,000 ppm.

Subsequently detailed geometallurgical analysis has been completed on Pakaka and Sessenge where the arsenic content has been modelled as part of the Mineral resource block model.

Metrics have been developed for stockpiling and blending, to dilute and minimise the impact of high arsenic in the overall plant feed. Additional work was carried out to identify the poor recovery related to the refractory component of the ore, while the pre-oxidation processes of the concentrate post ultrafine grinding was controlled or restricted to minimise arsenic mobilisation to solution.

18^th September 2018 Page 133

Kibali Gold Mine NI 43-101 Technical Report

14 Mineral Resource Estimates

14.1 Summary

Geological interpretation and Mineral Resource estimation were completed by Kibali with an effective date of 31^st December 2017.

Table 14-1 presents a summary of the Kibali Mineral Resource estimate, as of 31^st December 2017.

Table 14-1 Kibali Mine Mineral Resource Statement as of 31^st December 2017

Type	Category	Tonnes (Mt)	Grade (Au g/t)	Contained  Gold (Moz)	*Attributable Gold (Moz)
Stockpiles	Measured	1.7	1.45	0.080	0.036
Open Pits	Measured	8.6	2.63	0.73	0.33
	Indicated	39	2.11	2.6	1.2
	Inferred	22	1.8	1.3	0.59
Underground	Measured	12	5.57	2.1	0.96
	Indicated	65	3.64	7.6	3.4
	Inferred	22	2.8	2.0	0.91
Total Mineral Resources	Measured	22	4.11	3.0	1.3
	Indicated	104	3.07	10	4.6
	Measured and Indicated	126	3.26	13	5.9
	Inferred	44	2.3	3.3	1.5

*Attributable Gold (Moz) refers to the quantity attributable to Randgold based on Randgold’s 45% interest in the Kibali Goldmines. The Mineral Resource estimate has been prepared according to JORC (2012) Code. Kibali have reconciled the Mineral Resources to CIM (2014) Standards, and there are no material differences.

All Mineral Resources tabulations are reported inclusive of that material which is then modified to form Ore Reserves.

Open pit Mineral Resources are Mineral Resources within the $1,500/oz pit shell reported at an average cut-off grade of 0.6 g/t Au. Underground Mineral Resources in the KCD deposit are Mineral Resources, which meet a cut-off grade of 1.6 g/t Au and are reported insitu within a minimum mineable stope shape, at a gold price of $1,500/oz.

Mineral Resources were estimated by Simon Bottoms, CGeol, an officer of the company and Qualified Person.

Numbers may not add due to rounding.

Mineral Resources were estimated by Simon Bottoms, CGeol, an officer of the company and Qualified Person. The Kibali Project Mineral Resources consist of the KCD, Sessenge, Pakaka, Mengu Hill, Gorumbwa, Megi, Pamao, Kombokolo, Mengu Village, and Marakeke deposits. During 2017, only KCD, Sessenge, Pakaka, Gorumbwa, Pamao, and Kombokolo were updated, as a result of additional data from drilling, and/or updated geological mapping. No drilling has been completed on the Mengu Village and Marakeke deposits since their acquisition by Kibali Goldmines in 2009.

The Mineral Resource and Ore Reserve estimates have been prepared according to the guidelines Australasian Code for Reporting of Exploration Results, Mineral Resources and Ore Reserves standards and guidelines published and maintained by the Joint Ore Reserves Committee of the Australasian Institute of Mining and Metallurgy, Australian Institute of Geoscientists and Minerals Council of Australia (the JORC (2012) Code). Randgold has reconciled the Mineral Resources and Ore Reserves to Canadian Institute of Mining, Metallurgy

18^th September 2018 Page 134

Kibali Gold Mine NI 43-101 Technical Report

and Petroleum (CIM) 2014 Definition Standards for Mineral Resources and Mineral Reserves dated May 10, 2014 (CIM (2014) Standards) as incorporated with NI 43-101 and there are no material differences.

Definitions for resource categories used in this report are consistent with those defined by CIM (2014) and adopted by NI 43-101. In the CIM classification, a Mineral Resource is defined as “a concentration or occurrence of solid material of economic interest in or on the Earth’s crust in such form, grade or quality and quantity that there are reasonable prospects for eventual economic extraction”. Mineral Resources are classified into Measured, Indicated, and Inferred categories. A Mineral Reserve is defined as the “economically mineable part of a Measured and/or Indicated Mineral Resource” demonstrated by studies at Pre-Feasibility or Feasibility level as appropriate. Mineral Reserves are classified into Proved and Probable categories.

The estimation of Pakaka, KCD, Gorumbwa, Kombokolo, Pamao, and Sessenge models has been updated with all available drilling and workings as of the model cut-off date (Table 14-2). The models have been depleted using the December 2017 mined out shapes and surfaces. The KCD underground model update incorporates data from grade control and resource drilling up until 25th June 2017 for 3000 and 5000. For 9000 the 25th September 2017 cut-off was used, due to the amount of additional advanced grade control drilling completed and a review of the geological and mineralisation model from geological re-logging program.

Table 14-2 Summary of Deposits and Model Date

Deposit	Producing Status	Model Date
KCD UG	Active	25/09/2017
KCD OP	Active	25/11/2017
Sessenge	Unmined	25/08/2017
Gorumbwa	Unmined	30/06/2017
Pakaka	Active	15/09/2017
Kombokolo	Active	23/12/2017
Pamao	Unmined	04/07/2017
Mengu Hill	Depleted	24/10/2016
Mengu Village	Unmined	24/01/2008
Marakeke	Unmined	24/01/2008

Models for active producing mines were updated on a monthly basis to incorporate all additional grade control drilling results throughout 2017 (Table 14-2). Sessenge was updated in August 2017 with grade control infill drilling prior to commencement of the scheduled mining at the beginning of 2018. The Gorumbwa deposit has been updated using drilling and updated void shapes which was aimed at re-defining the potential resources. Pamao was updated following a geological review and to identify potential upside for further resource definition drilling in 2018.

For the Mengu Village and Marakeke deposits, the models are inherited resources and have not yet been updated since their acquisition by Kibali Goldmines in 2009. No drilling has been completed on these deposits since their acquisition by Kibali Goldmines in 2009.

The cut-off grade selected for limiting each of the Mineral Resources corresponds to the insitu marginal cut-off grade using a gold price of $1,500/oz.

18^th September 2018 Page 135

Kibali Gold Mine NI 43-101 Technical Report

For the open pit Mineral Resources, the pit shell selected for limiting each of the Mineral Resources corresponds to a gold price of $1,500/oz. As a result of the optimisation process, this pit shell selection will result in the highest undiscounted net present value of the deposit, at $1,500/oz.

Underground Mineral Resources were reported within a minimum mineable stope shape, applying reasonable mineability constraints, including a minimum mining width, a reasonable distance from current or planned development, and a measure of assumed profitability at the related resource cut-off grade, thus deemed as having a reasonable prospect of eventual economic extraction.

The Qualified Person is not aware of any environmental, permitting, legal, title, socioeconomic, marketing, metallurgical, fiscal, or other relevant factors that could materially affect the Mineral Resource estimate.

18^th September 2018 Page 136

Kibali Gold Mine NI 43-101 Technical Report

14.2 Resource Database

KCD

During 2017, a total of 994 drill holes were completed for a sum of 12,127 m. This drilling in KCD consists of combination of grade control drilling at 10 m by 5 m spacing within high-grade shoot zones, 20 m by 5 m spacing within low-grade zones and 20 m by 20 m spacing in waste zones for open pit and approximately 20 m by 15 m spaced diamond drilling for underground grade control. A summary of KCD data used for the updated Mineral Resource estimate is presented in Table 14-3.

Table 14-3 Drill Summary of KCD Used in 2017 Mineral Resource Estimate

Company	Year Completed	Drill Type	Number of  Holes	Minimum  Depth (m)	Maximum  Depth (m)	Total Drilled (m)
Kibali Goldmines	2017	DDH	492	6	1,491	84,702
Kibali Goldmines		GT (Geotech DDH)	8	100	280	1,073
Kibali Goldmines		RC	494	12	150	26,352
Kibali Goldmines	2016	DDH	502	9.24	565	64,670
Kibali Goldmines		GT	14	66	427	3,227
Kibali Goldmines	2015	DDH	250	45	464	41,793
Kibali Goldmines		GT	18	57	230	2,698
Kibali Goldmines		RC	858	6	100	35,122
Kibali Goldmines	2014	DDH	101	26.3	800	15,165
Kibali Goldmines		GT	8	135	321	1,894
Kibali Goldmines		RC	1,743	6	93	83,439
Kibali Goldmines	2013	DDH	28	16.3	801	12,534
Kibali Goldmines		GT	3	194.6	723	1,364
Kibali Goldmines		RC	1307	12	99	66,668
Kibali Goldmines	2012	DDH	22	102	1,092	10,761
Kibali Goldmines		GT	11	40	801	2,765
Kibali Goldmines		RC	1,763	3	100	91,369
Kibali Goldmines	2011	DDH	15	8.95	1,347	4,991
Kibali Goldmines		GT	69	10.8	860	10,157
Kibali Goldmines		RC	1,720	4	110	51,258
Kibali Goldmines	2010	DDH	58	25.1	942	27,166
Kibali Goldmines		GT	11	101.2	728	5,261
Kibali Goldmines		RC	65	6	110	3,484
Kibali Goldmines	2009	DDH	9	72.3	798	2,938
Moto		DDH	67	71.2	790	23,035
Moto	2008	DDH	98	52.89	861	50,741
Moto		GT	11	403.35	650	5,954
Moto	2007	DDH	67	190.85	953	40,982
Moto		GT	6	170	420	1,871
Moto	2006	DDH	111	23	699	34,704
Moto		GT	2	230	263	493
Moto	2005	DDH	51	116	666	14,890
Moto		RC	10	112	120	1,192
Moto	2004	DDH	9	149.7	421	1,904
Moto		RC	34	31	100	2,596
Total			10,035			829,211

18^th September 2018 Page 137

Kibali Gold Mine NI 43-101 Technical Report

KCD Open Pit

Table 14-4 presents statistics of the composite samples used in the 2017 Mineral Resource estimate for KCD Open Pit. There has been a 10% increase in data points and 5% drop in capped mean grade for the 2017 vs 2016 open pit model, of which most of the additional data has come from grade control drilling.

Table 14-4 KCD Open Pit Resources Composite Data – 2017 Mineral Resource Estimate

Domain	No of Samples	Minimum Au (g/t)	Maximum Au (g/t)	Mean Au  (g/t)	CV Au	Capped  Au (g/t)	Mean Capped  Au (g/t)	CV Capped  Au	No of Samples Capped
3000	38,647	0.01	499.70	1.75	2.89	79.00	1.71	2.34	64
3100	7,317	0.01	514.37	7.56	2.00	100.00	7.17	1.52	80
5000	25,042	0.01	260.00	1.79	3.06	70.00	1.73	2.40	32
5100	9,084	0.01	3,008.00	6.82	5.17	100.00	6.21	1.33	18
9000	2,255	0.01	38.20	0.98	2.08	16.11	0.95	1.84	7
9100	33	0.05	7.23	2.98	0.65	7.23	2.98	0.65	-
Total	88,412	0.05	3,008.00	2.67	4.90	100.00	2.55	2.29	201

KCD Underground

3000 and 5000 Domains

Table 14-5 presents statistics of the composite samples used in the 2017 Mineral Resource estimate for KCD underground 3000 and 5000 Domains. Since the 2016 resource model, global 3000 and 5000 sample data have increased 52%, mainly as a result of infill grade control drilling in 5000 and advanced grade control drilling in 3000 lode. The capped mean grade has increased by 11%.

Table 14-5 KCD 3000 and 5000 Underground Resources Composite Data – 2017 Mineral Resource Estimate

Domain	No of Samples	Minimum Au (g/t)	Maximum Au (g/t)	Mean Au (g/t)	CV Au	Capped  Au (g/t)	Mean Capped  Au (g/t)	CV Capped  Au	No of Samples Capped
3000	6,158	0.005	112.0	0.77	3.32	41.00	0.70	2.23	29
3100	3,876	0.005	123.2	3.65	1.76	80.00	3.55	1.59	26
5000	14,555	0.005	433.6	1.03	4.22	70.00	0.98	2.41	18
5100	16,419	0.006	540.0	6.98	1.58	100.00	6.81	1.22	46
Total	41,008	0.005	540.0	3.62	2.79	80.00	3.52	1.83	119

9000 Domain

Table 14-6 presents statistics of the composite samples used in the 2017 Mineral Resource estimate for KCD underground 9000 Domain.

18^th September 2018 Page 138

Kibali Gold Mine NI 43-101 Technical Report

Table 14-6 KCD 9000 Underground Mineral Resource Estimates Composite Data – 2017 Mineral Resource

Estimate

Domain	No of Samples	Minimum Au (g/t)	Maximum Au (g/t)	Mean Au (g/t)	CV Au	Capped  Au (g/t)	Mean Capped  Au (g/t)	CV Capped  Au	No of Samples Capped
9000	21,496	0.005	72.9	0.67	2.98	17	0.63	2.28	58
9100	7,823	0.005	200.7	5.49	1.43	50	5.34	1.23	60
Total	29,319	0.005	200.7	1.95	2.56	50	1.89	2.00	118

Since the 2016 resource model there has been an increase of 83% in the number of samples in the 9000 underground domains, the majority of which are low-grade samples in the low-grade halo. The high-grade domain data has increased by approximately 50% from advanced grade control drilling, with the mean capped grade dropping from 6.0 g/t Au to 5.3 g/t Au.

Sessenge

Between 2016 and 2017, additional advanced grade control drilling was completed for 296 holes for 19,109 m, of which three holes were DDH to be used for metallurgical testwork. Table 14-7 summarises the drill holes used in the 2017 Sessenge estimation.

Table 14-7 Sessenge Drill Summary of Holes Used in 2017 Mineral Resource Estimate

Company	Year Completed	Drill Type	Number of  Holes	Minimum  Depth (m)	Maximum  Depth (m)	Total Drilled (m)
Kibali Goldmines	2017	DDH	3	161	245	613
Kibali Goldmines	2017	RC	293	12	177	18,496
Kibali Goldmines	2016	DDH	7	74	194	831
Kibali Goldmines	2016	RC	207	19	160	11,620
Kibali Goldmines	2015	DDH	1	152	152	152
Kibali Goldmines	2015	RC	108	40	130	6,609
Kibali Goldmines	2011	GT	5	30	31	152
Kibali Goldmines	2010	DDH	6	120	300	1,237
Kibali Goldmines	2010	RC	160	15	150	8,344
Moto	2008	DDH	1	266	266	266
Moto	2008	GT	3	151	157	458
Moto	2006	DDH	15	39	353	2,911
Moto	2006	RC	23	82	100	2,282
Moto	2005	DDH	13	158	249	2,521
Moto	2005	RC	87	40	160	8,288
Moto	2004	RC	41	50	60	2,210
Total			973			66,990

During 2017, Sessenge was remodelled based on an updated folded BIF model interpretation. This resulted in an increase in the global mean grade from 1.97 g/t to 2.09 g/t Au and significant transfer of samples from 1001 to 1002 domain. In addition, two high-grade shoots were modelled with mean grades above 4 g/t Au.

Table 14-8 presents statistics of the composite samples used in the 2017 Mineral Resource estimate for Sessenge.

18^th September 2018 Page 139

Kibali Gold Mine NI 43-101 Technical Report

Table 14-8 Sessenge Composite Data – 2017 Mineral Resource Estimate

Domain	No of Samples	Minimum Au (g/t)	Maximum Au (g/t)	Mean Au (g/t)	CV Au	Capped Au (g/t)	Mean Capped Au (g/t)	CV Capped Au	No of Samples Capped
9001	126	0.05	9.98	2.21	0.81	6.5	2.15	0.74	4
9002	4704	0.02	45	1.54	1.46	25	1.53	1.38	3
9003	309	0.005	18.81	1.51	1.37	9	1.43	1.1	4
9007	419	0.005	5.71	1.19	0.83	5.71	1.19	0.83	0
9008	478	0.005	26	1.95	1.64	20	1.92	1.57	4
9009	272	0.02	19.6	1.33	1.5	9	1.25	1.2	4
9102	771	0.03	31	3.93	0.81	19	3.9	0.78	2
9103	728	0.05	21.3	4.98	0.63	21.3	4.98	0.63	0
Total	7807	0.005	45	2.11	1.29	25	2.09	1.24	21

Gorumbwa

Table 14-9 summarises the drill holes used in the 2017 Gorumbwa estimation.

Table 14-9 Drill Summary of Gorumbwa Holes Used in 2017 Mineral Resource Estimate

Year Completed	Drill Type	No of Holes	Minimum Depth (m)	Maximum Depth (m)	Total Drilled (m)	Mean Depth (m)
2017	RC	205	18	177	9354	46
2016	RC with DD Tail	34	104	224.0	6,226.0	178.0
	RC	195	23	245.0	23,502.0	121.0
	DDH	15	98	225.0	2,616.0	174.0
2015	DDH	31	53	407.0	4,835.0	156.0
	RC	152	20	120.0	8,716.0	57.0
2014	RC	54	12	204.0	4,480.0	83.0
	RC with DD Tail	3	125	178.0	428.0	143.0
	GT	6	60	279.0	1,074.0	179.0
	DDH	66	44	66.0	14,562.0	221.0
	Trench	1	159	159.0	159.0	159.0
2012	DDH	12	63	615.0	3,414.0	284.0
2011	DDH	1	900	900.0	900.0	900.0
2006	RC	22	61	180.0	3,473.0	158.0
	Geotech	1	150	150.0	150.0	150.0
2005	RC	53	34	180.0	5,545.0	105.0
	DDH	2	41	300.0	341.0	170.0
2004	RC	7	50	64.0	374.0	53.0
	DDH	25	60	400.0	5,045.0	202.0
1996	DDH	3	207	395.0	907.0	302.0
Total		888	12	900	96,100	107

The Gorumbwa deposit was mined between the 1950s and the 1990s. Additional drilling completed by Kibali in 2012 provided some varying opinions about the mined-out shapes and the potential for further mineralisation which may have been ‘missed’ by historic mining. Between 2014 and 2015 additional holes were drilled to redefine the mined-out shapes and closely spaced grade control orientation drilling was also completed. The 2016 drilling program was targeted to confirm the void models and to test the possibility of additional ore tonnes at depth. The 2017 drilling program aimed to convert the remaining Inferred materials to Indicated and test the continuity of 1004 underneath of the open pit mined old mined shape. Data used for 2017 model update is last updated on 30^th June 2017.

18^th September 2018 Page 140

Kibali Gold Mine NI 43-101 Technical Report

Table 14-10 presents statistics of the composite samples used in the 2017 Mineral Resource estimate for Gorumbwa.

Table 14-10 Gorumbwa Composite Data – 2017 Mineral Resource Estimate

Domain	No of Samples	Minimum Au (g/t)	Maximum Au (g/t)	Mean Au (g/t)	CV Au	Capped Au (g/ t)	Mean Capped Au  (g/t)	CV Capped  Au	No of Samples Capped
1001	1,512	0.005	36.2	2.4	1.55	22	2.36	1.43	13
1002	419	0.019	12.7	1.37	1.06	12.7	1.37	1.04	0
1003	549	0.005	240	2.15	4.87	17	1.7	1.46	2
1004	2,757	0.005	174.98	3.49	2.47	33	3.13	1.7	33
1005	93	0.03	46.4	2.65	2.42	14	2.03	1.33	2
1006	759	0.025	94.2	2.22	2.49	21	1.98	1.67	7
1007	215	0.005	91.3	5.01	2.2	26	4.14	1.61	7
1008	636	0.03	22.7	1.80	1.43	22.7	1.80	1.41	0
1009	107	0.05	16.35	1.56	1.57	8	1.42	1.23	2
1010	72	0.04	27.6	3.41	1.61	15	2.98	1.46	4
1011	155	0.015	4.34	0.7	0.97	4.34	0.7	0.95	0
1012	57	0.22	45.9	4.2	1.95	15	3.3	1.3	3
Total	7,331	0.01	240	2.71	2.57	33	2.47	1.69	73

Pakaka

The Mineral Resource estimate at Pakaka, an operating mine, was updated using additional RC drilling completed for grade control and Advanced Grade Control. These RC samples have also been used to increase the confidence of the geometallurgical model, due to the presence of Arsenic. A total of 50,317 m of additional RC drilling was completed since the 2016 resource model and has been used to update the model. Advanced grade control holes were drilled on a 40 m by 10 m spacing and infill grade control drilling was completed at 20 m by 5 m spacing which is the optimum drill spacing for this style of mineralisation and its extent as confirmed by an orientated drill campaign completed in late 2015.

A summary of the historical and recent drilling undertaken at Pakaka is presented in Table 14-11.

Table 14-11 Drill Summary of Pakaka Holes Used in 2017 Mineral Resource Estimate

Company	Year Completed	Drill type	No of Holes	Minimum Depth	Maximum Depth	Total Drilled
Kibali Goldmines	2017	RC	403	30	213	50,317
Kibali Goldmines	2016	RC	1,145	15	182	84,660
Kibali Goldmines	2015	DDH	5	103	143	617
Kibali Goldmines	2015	RC	690	6	131	38,180
Kibali Goldmines	2014	RC	2	100	150	250
Kibali Goldmines	2013	DDH	6	97	295	1,075
Kibali Goldmines	2012	DDH	9	26	700	1,921
Kibali Goldmines	2012	GT	5	171	323	1,161
Kibali Goldmines	2011	DDH	1	700	700	700
Kibali Goldmines	2011	GT	1	169	169	169
Kibali Goldmines	2011	RC	5	100	150	550
Kibali Goldmines	2010	RC	3	100	100	300
Moto	2007	DDH	10	341	449	3,910
Moto	2007	RC	5	48	151	504
Moto	2006	DDH	41	45	182	4,475
Moto	2006	GT	5	80	191	692
Moto	2006	RC	7	45	80	415
Moto	2005	DDH	73	100	347	15,434

18^th September 2018 Page 141

Kibali Gold Mine NI 43-101 Technical Report

Company	Year Completed	Drill type	No of Holes	Minimum Depth	Maximum Depth	Total Drilled
Moto	2005	RC	34	40	140	3,279
Moto	2004	DDH	16	120	230	2,892
Moto	2004	RC	159	30	130	10,679
Barrick	1996	DDH	8	84	188	920
Total			2,633			223,100

Table 14-12 presents statistics of the composite samples used in the 2017 Mineral Resource estimate for Pakaka. A comparison of the drilling data used in the 2017 estimate versus 2016 resource estimate shows that a significant number of infill samples have been added and the mean grade has remained stable.

Table 14-12 Pakaka Composite Data – 2017 Mineral Resource Estimate

Domain	No of Samples	Minimum Au (g/t)	Maximum Au (g/t)	Mean Au  (g/t )	CV Au	Capped Au (g/t)	Mean Capped Au (g/t)	CV Capped Au	No of Samples  Capped
1001	17,140	0.01	65.42	1.5	1.44	32	1.49	1.36	13
1007	163	0.03	6.99	1.23	1	6.99	1.23	0.98	0
1101	2,613	0.01	89.43	7.02	0.96	34	6.93	0.87	11
1102	101	0.37	520.00	10.4	5.06	13	4.36	0.74	8
1103	199	0.79	24.60	3.97	0.84	10	3.7	0.62	9
1105	1,141	0.01	60.00	4.13	1.2	21	3.89	0.82	19
Total	21,357	0.01	520.00	2.37	2.19	34	2.31	1.47	60

Kombokolo

Table 14-13 summarises the exploration that has been undertaken on Kombokolo since 2005.

Table 14-13 Drill Summary of Kombokolo Holes Used in December 2017 Mineral Resource Estimate

Company	Year Completed	Drill type	No of Holes	Minimum Depth	Maximum Depth	Total Drilled
Kibali Goldmines	2017	RC	820	12	189	63,182
Kibali Goldmines	2017	DDH	36	88	301	6,135
Kibali Goldmines	2016	Trench	30	16	174	2,121
Kibali Goldmines	2016	PIT	64	1	10	210
Kibali Goldmines	2015	GT	1	161	161	161
Kibali Goldmines	2015	Trench	1	200	200	200
Kibali Goldmines	2014	Trench	8	42	290	928
Kibali Goldmines	2014	GT	1	86	86	86
Moto	2006	DDH	2	100	148	248
Moto	2006	GT	1	90	90	90
Moto	2005	RC	30	59	170	3,382
Total			994	78	165	76,743

The Kombokolo deposit is comprised of nine lodes, seven of which are low-grade domains (1001 to 1007) and two are high-grade domains (1101 and 1102). The high-grade domains 1101 and 1102 were identified post 2017 audit model using specific geological criteria observed in DDH ore and are hosted within 1001 and 1002 respectively. The two main lodes are 1001 and 1002 and are generally separated by an internal sericite schist waste unit. In some cases, the waste between the lodes is the same geology as the resource estimate but is uneconomic. The model had been updated as a result of drilling, pitting, and trenching completed in 2017 to confirm the extent of mineralisation and the historical works completed.

18^th September 2018 Page 142

Kibali Gold Mine NI 43-101 Technical Report

Table 14-14 presents statistics of the composite samples used in the 2017 Mineral Resource estimate for Kombokolo. Since the 2016 resource model a total of 820 holes were added to the Kombokolo resource model, the majority of which are in the main low-grade domains 1001 and 1002. Comparison of the composite data shows the global mean grade has increased along with the capped gold mean grade, in addition to the increase in number of capped samples with a lower top cutting value. With the generation of high-grade domains, the CV of both raw and capped data has decrease, indicating an improved distribution within the updated domains.

Table 14-14 Kombokolo Composite Data – 2017 Mineral Resource Estimate

Domain	No of Samples	Minimum Au (g/t)	Maximum Au (g/t)	Mean Au (g/t)	CV Au	Capped Au (g/t)	Mean Capped Au (g/t)	CV Capped Au	No of Samples Capped
1001	4,071	0.005	79.7	3.11	1.77	39.6	3.01	1.56	21
1002	2,181	0.01	124.4	2.95	1.71	18.7	2.74	1.21	32
1003	57	0.02	5.76	1.26	0.93	5.76	1.26	0.92	-
1004	15	0.34	3.43	0.97	0.84	3.43	0.97	0.82	-
1005	33	0.005	9.88	2.61	1.13	9.88	2.61	1.13	-
1006	5	0.5	3.91	1.8	0.77	3.91	1.80	0.78	-
1007	2	1.18	2.63	2.08	0.54	2.63	2.08	0.49	-
Total	6,364	0.005	124.40	3.03	1.75	39.60	2.89	1.46	53

The log histogram combining all data presents a fairy good log normal distribution showing no bimodal population despite the presence of the high-grade envelops which contribute at 35% of the data (above 2.5 g/t Au). The log probability plot shows quite good homogeneity apart from the extremities of the high-grade tails.

Pamao

The Pamao Mineral Resource estimate was updated in 2017, the first time since 2010, with the addition of 62 holes (two are DDH and 60 are RC). The mineralisation model was fully revised with significant changes in interpretation resulting in an increase in global mean grade from the removal of waste samples from the low-grade wireframes. The Pamao deposit comprises three different lodes named as 2001 at the top, 2002 which represents the main zone and 2003, below. The lodes are generally separated by an internal waste within the meta sandstone unit.

A summary of the historical and recent drilling undertaken at Pamao is presented in Table 14-15.

18^th September 2018 Page 143

Kibali Gold Mine NI 43-101 Technical Report

Table 14-15 Drill Summary of Pamao Holes Used in 2017 Mineral Resource Estimate

Company	Year Completed	Drill Hole Type	No of Holes	Min Depth	Max Depth	Total Drilled
Kibali Goldmines	2017	DDH	2	108	174	282
Kibali Goldmines	2017	RC	60	40	66	2,712
Kibali Goldmines	2016	DDH	7	58	189	845
Moto	2007	GT	4	80	105	385
Moto	2005	DDH	24	81	200	3,495
Moto	2005	RC	26	60	130	2,640
Moto	2004	RC	118	40	160	7,920
Total			241	40	200	18,280

Table 14-16 presents statistics of the composite samples used in the 2017 Mineral Resource estimate for Pamao.

Table 14-16 Pamao Composite Data – 2017 Mineral Resource Estimate

Domain	No of Samples	Minimum Au (g/t)	Maximum Au (g/t)	Mean Au (g/t)	CV Au	Capped Au (g/t)	Mean Capped Au (g/t)	CV Capped Au	No of Samples Capped
2001	186	0.01	3.67	0.8	0.8	3.67	0.8	0.79	0
2002	1,348	0.01	26.8	1.4	1.41	26.8	1.4	1.41	0
2003	648	0.01	21.4	1.26	1.4	21.4	1.26	1.39	0
Total	2,182	0.01	26.8	1.3	1.41	26.8	1.3	1.4	0

Mengu Hill

The Mengu Hill resource model was not updated during 2017. The 2016 resource model used a database named ‘mgh_res_2016_10_24.accdb’ which was used for the Mengu estimation. Between 2014 and 2016, grade control together with advanced grade control drill holes have been completed which were used to update the grade control model for mining. A total of 14,246 m of RC drilling was completed in 2016 which substantially increased the confidence in the Mengu model. Additional works completed in 2016 shows Mengu to have increased in both resources and reserves, thereby increasing the expected life of the pit by five months.

A summary of the historical and recent drilling undertaken at Mengu Hill is presented in Table 14-17.

18^th September 2018 Page 144

Kibali Gold Mine NI 43-101 Technical Report

Table 14-17 Mengu Drill Summary of Mengu Holes Used in 2016 Mineral Resource Estimate

Company	Year Completed	Drill Type	Number of Holes	Minimum Depth (m)	Maximum Depth (m)	Total Drilled (m)
Kibali Goldmines	2016	Pit	21	1.7	14.0	101.6
Kibali Goldmines		RC	307	6.0	112.0	14,246.0
Kibali Goldmines	2015	DDH	1	806.4	806.4	806.4
Kibali Goldmines		Pit	31	1.0	3.8	62.6
Kibali Goldmines		RC	444	12.0	84.0	16,428.0
Kibali Goldmines		RC with DD Tail	17	78.0	332.3	2,714.6
Kibali Goldmines	2014	Pit	10	2.1	2.8	25.2
Kibali Goldmines		RC	1,011	8.0	175.0	42,160.0
Kibali Goldmines		Trench	2	62.0	150.0	212
Kibali Goldmines	2013	DDH	22	39.5	330.2	2,620.8
Kibali Goldmines		RC	1	200.0	200.0	200.0
Kibali Goldmines	2012	DDH	9	152.8	395.3	2,348.1
Kibali Goldmines		Geotech	11	9.9	266.1	1,105.2
Kibali Goldmines		Pit	18	1.8	4.3	45.1
Moto	2006	DDH	1	85.0	85.0	85.0
Moto		Geotech	3	150.2	180.0	480.4
Moto	2005	DDH	37	53.0	260.0	5,763.5
Moto		RC	58	40.0	240.0	4,879.0
Moto	2004	RC	97	60.0	60.0	5,820.0
Total			2,101.00			10,0103.4

Table 14-18 presents statistics of the composite samples used in the 2016 Mineral Resource estimate for Mengu Hill.

Table 14-18 Mengu Hill Composite Data – 2016 Mineral Resource Estimate

Domain	No of Samples	Minimum Au (g/t)	Maximum Au (g/t)	Mean Au (g/t)	CV Au	Capped Au (g/t)	Mean Capped Au (g/t)	CV Capped Au	No of Samples Capped
1001	9,467	0.01	33.60	1.14	1.43	30.00	1.14	1.42	1
1002	125	0.03	4.40	1.19	0.95	4.40	1.19	0.91	0
1101	7,587	0.02	117.00	5.76	1.19	65.00	5.74	1.14	5
1102	25	1.00	31.10	6.69	1.10	10.00	4.89	0.58	4
1103	139	0.05	11.30	3.16	0.69	11.30	3.16	0.67	0
Total	17,343	0.01	117.00	3.20	1.64		3.19	1.58	10

Mengu Village and Marakeke

The drilling database for Mengu Village consists of 51 RC drill holes, totalling 3,702 m, drilled by Moto in 2004-2005. The drilling database for Marakeke consists of 72 RC drill holes, totalling 4,957 m, drilled by Moto in 2004 to 2005.

No additional drilling has taken place a Mengu Village or Marakeke since 2005.

18^th September 2018 Page 145

Kibali Gold Mine NI 43-101 Technical Report

14.3 Geological Modelling

Geological interpretation and modelling are based on the following standard procedures:

☐ Geological cross sections and long sections are generated and updated during any drill campaigns. These are then scanned and georeferenced to be used as a basis for 3D modelling.

☐ Geological interpretations are digitised as polylines on cross sections spaced 10 m apart on KCD. Geology, alteration, and low and high-grade polylines are snapped on each section to the corresponding sample interval. In areas of complex folding additional polylines are wireframed between sections to build a valid 3D solid.

☐ Mineralisation domains are sub domained into low-grade and high-grade domains, utilising contact analysis and domain stationarity tests.

☐ The geological and mineralisation models are updated monthly, and quarterly when additional grade control data is available.

☐ Interpretations are regularly cross checked with DD core and RC chips to ensure the model is representative.

Statistical and visual analysis of the data showed that a suitable geological related cut-off grade was approximately 0.5 g/t Au for the KCD and Sessenge deposits. For the Gorumbwa Pakaka, Kombokolo, and Pamao deposits the ore and waste contacts are also modelled around 0.5 g/t Au. The resulting low-grade mineralised envelopes incorporate minor amounts of internal sub-grade material content to preserve continuity. During interpretation, efforts were made to minimise the amount of sub-grade material included within each of the lode wireframes.

Mineralisation domains were built with a combination of grade, lithology, alteration, structures, and the presence of pyrite content. In areas where further high-grade shoots are evident, high-grade continuity wireframes were also considered.

The intention of the geological domaining is to generate a single stationary geostatistical population for each of the domains. If this was not possible, then these areas were sub-divided into sub-domains thereby ensuring that single populations were created. Boundary analysis (Figure 14-1) is completed to check if there is a sharp change in grade profile across a domain boundary. This helps delineate the rod-like high-grade mineralisation shoots noted in the KCD, Sessenge, Kombokolo, and Pakaka deposits.

18^th September 2018 Page 146

Kibali Gold Mine NI 43-101 Technical Report

Figure 14-1 Boundary Analysis between HG (5101) and LG (5005) Domains from KCD

The drill holes are coded by each domain. These codes are used for statistical analysis and domain control during the estimation process. The coding of the drill holes and block model is prioritised to ensure that the high-grade domain codes are preserved, when they are situated within surrounding low grade mineralisation envelopes. All drill holes are composited to 2 m, to create homogenous sample weights within the mineralised zone for estimation.

To ensure consistency of the domaining controls used, the database and geological block model are both flagged with the same codes defining the mineralised envelopes that a particular composite falls within. The high-grade mineralised envelopes will predominantly situate within low grade mineralisation wireframes, which are built independently of each other. Due to the fact that boolean operations are not utilised to remove these overlaps between internal high grade shoot models and surrounding low grade mineralisation envelope wireframes, care is taken to avoid the double-counting of samples and blocks.

KCD and Sessenge

KCD areas have been selectively modelled using a combination of grade continuity, alteration, mineralisation, and structural readings, where available.

KCD Mineralisation Modelling

Broad mineralisation lodes were modelled using all available drilling data and geological information such as lithology, alteration, structural, and mineralisation characteristics of the ore zones. These mineralisation wireframes generally followed a cut-off grade of 0.5 g/t Au, but did include some internal dilution, where applicable, in order to create reasonably continuous envelopes. Some of these broad envelopes often contain multiple rod-like high-grade zones within them. These high-grade zones are associated with strong ACSA alteration and fine-grained disseminated pyrite and have been modelled separately at a nominal cut-off grade of 2.5

18^th September 2018 Page 147

Kibali Gold Mine NI 43-101 Technical Report

g/t Au. To maintain continuity, where applicable, sections of the lower grade material were included as internal dilution. The strong ACSA alteration and pyrite zones were also modelled.

The mineralisation and lithology models are generated from geo-referenced paper cross sections into Maptek Vulcan 3D software. Strings are generated on 10 m spaced vertical sections in a NE direction across the extent of the mineralisation. Wireframes have been snapped, where possible, to the drill hole sample intervals to create a precise boundary between the mineralisation of the low-grade and high-grade envelopes. The resulting interpretation produces consistent geometry and geological continuity for the plunging mineralised lodes.

Mineralisation at KCD and Sessenge has been regrouped into three lodes (3000, 5000 and 9000). The 9000 lode extends up plunge to surface at Sessenge, enabling the two deposits areas to be joined together into a single unit (Figure 14-2). The 3000 lode comprises most of the near surface mineralisation thereby making up most the open pit resources, with current work on-going to turn the down plunge extension into an underground operation. The 5000 contains the majority of the mineralisation and is the dominant source of ore from underground.

Thin continuous barren intrusive/volcanic dolerite units are interspersed within the metasediments units and have been used as marker units during geological interpretation. These barren units have been modelled independently and flagged as code 800 to both the composites and the block model. To preserve the overall continuity of the mineralisation, wireframes were modelled not only taking grade into account, but, they also used various geological constraints including the intrusive/barren structures.

Figure 14-2 3D View of KCD Sessenge Mineralisation Lodes (3000 Lodes in Orange, 5000 in Red, 9000 in

Pink) and $1,000 Pit Design in Grey

18^th September 2018 Page 148

Kibali Gold Mine NI 43-101 Technical Report

Structure, Alteration and Mineralisation

Structural analysis and re-logging of the core, combined with cross-section construction, has provided information to reasonably understand the location of ACSA alteration and mineralisation. The following key observations can be made:

☐ Stereographic projections of structural data acquired from the core indicate that the axes of folds shown on the cross-sections are approximately parallel to the overall plunge of the mineralised shoots and lineation as measured in the open pit (Allibone, 2015). This is consistent with an intimate relationship between folding and later ore shoot development.

☐ Alteration and mineralisation are spatially related to banded iron formation (BIF). Wherever the mineralisation fluids intersect BIF, this appears to have promoted the deposition of gold.

☐ Alteration also preferentially extends beyond the margins of BIF, following axial planes of some of the major fold hinges, or along the preferred structural grain of the sericite alteration.

☐ Alteration and mineralisation occur preferentially in the footwall rock immediately below the major structural break that separates the chert and carbonaceous shale-bearing sedimentary package from underlying rocks.

Based on these observations, and because of the distribution and fold shape of BIF, modelling the continuity and morphology of the lodes has been greatly improved. Folding, with alteration and mineralisation following the main fold plunge direction, can be traced with confidence in the same structural setting from one cross-section to the next.

Like the host rocks in the 5000 to 9000 lodes, the most striking feature of the 3000 lode host rocks is the pervasive sericitic foliation. This foliation is subparallel to the fold axial surfaces of the mapped folds. The axial surfaces of the folds in the 3000 lodes are subparallel to the 5000 to 9000 axial surfaces although these generally become steeper and locally more curved (due to subsequent deformation) down section in the 5000 to 9000 lodes. The similar orientation, form and relative paragenetic timing of the sericitic foliation in both areas suggests that the associated folds in both areas are the same generation and not fundamentally different in terms of timing or original orientation.

ACSA-A alteration is controlled by the fold axial surface foliation in both areas and is most intense in subparallel shears. The overprinting ACSA-B alteration and mineralisation is significantly different in the 3000 lodes but this is interpreted as the result of local host rock variation rather than a significant change in the mineralising hydrothermal fluids. Mineralisation in the 3000 lode is hosted primarily in brecciated cherts and BIF in the hinges of folds, whereas mineralisation in the 5000 to 9000 lodes is hosted predominantly in the hinge zones and limbs of folded BIF units. Due to the predominance of chert in the 3000 lodes, the pyrite-pyrrhotite assemblage is more common than the pyrite dominated 5000 to 9000 lodes that are hosted in BIF.

Mineralised zones in the 3000 lodes are typically hosted in tightly folded hinge zones where relatively brittle host rocks (cherts and BIF) have resisted folding and been brecciated and sheared. The 3000 lode contains abundant cm to m scale early dykes and sills that crosscut these units and, although not particularly mineralised, have been the focus of strong shearing

18^th September 2018 Page 149

Kibali Gold Mine NI 43-101 Technical Report

and ACSA-A alteration. This focussed shearing has strongly affected the host rocks and cherts and BIF are typically brecciated and mineralised along the contacts with these altered, rheologically weak, units.

Shearing has also focussed on the carbonaceous phyllite units and carbonaceous shears typically contain lenticular clasts of other sheared lithologies including siltstone, chert, and igneous clasts. Clasts are commonly replaced by pyrite ± pyrrhotite that commonly carry elevated gold values, but mineralised sections are volumetrically minor in the 3000 lode. The best mineralised areas are hosted in sheared and brecciated cherts and BIFs on the margins of such zones.

Each of the mineralisation wireframe is modelled separately and snapped to drill hole where applicable. Lithological, weathering, and redox wireframes also modelled from drilled hole data which were flagged into the database and the block model with their respective priorities.

Gorumbwa

The mineralisation outlines within the Gorumbwa areas have been selectively modelled using a combination of grade continuity, alteration, mineralisation, and structural readings where available.

Mineralisation at Gorumbwa is hosted almost exclusively within the meta sandstone unit, with minor sporadic mineralisation noted in a conglomerate unit that occurs beneath the metasandstone. Mineralisation is divided into twelve lodes (1001 to 1012) (Figure 14-3) which coarsely trend west to west-SW, dip to the NW, and plunge to the ENE at approximately 30°. The lenses are echelon like in vertical stacking, with only the main 1004 lens being the most consistent in continuity. The stratigraphically higher upper lenses include 1001 to 1003, with 1005 to 1012 the deeper footwall lenses. Higher grade within the lodes tends to the central area of the shoots where a higher strain environment increased accommodation space and hydrothermal fluid inflows. Historic mining focussed on the extraction of the main (1004) lode. The styles of mineralisation vary from KCD, with the dominant style being moderate to strong silicification and sericitisation with minimal pyrite, with a low correlation between sulphide and gold content. The second style is the typical ‘ACSA’ style noted at KCD where the gold is proportional to pyrite percentages, though the iron carbonate is predominantly ankerite, unlike KCD which is dominated by siderite. This style is mainly observed in the main 1004 lode. The third style is visible gold within late, moderately to strongly silicification.

Mineralisation is structurally controlled within a NE trending corridor where the S1 foliation strikes east-west. The corridor is bounded by NE crosscutting structures on the eastern and western margins. These NE structures are indicated by the discontinuity of the red pebble conglomerate horizon in the west, micro folding within the lithological units near to the structures and rotation of S1 foliation laterally across the 200 m wide mineralised area. The structures may be ductile. Refolded folds are not observed but may have been seen in the high-grade shoot which was mined and makes up the void area. Each of the mineralisation wireframe is modelled separately and snapped to drill hole where applicable. Lithological, weathering, and redox wireframes also

18^th September 2018 Page 150

Kibali Gold Mine NI 43-101 Technical Report

modelled from drilled hole data which were flagged into the database and the block model with their respective.

Historical mining in these areas has resulted in the high-grade core being mined out and thus all of the current lodes have been classified as low-grade envelopes.

Figure 14-3 3D View of Gorumbwa Mineralisation (Yellow) Depletion Model (Magenta) and $1,000 & $1,500

whittle Pit Shells

Pakaka

Gold mineralisation at Pakaka-Pamao is hosted by volcano-sedimentary conglomerate interbedded with minor tuffaceous units. Significantly, ironstone units are rare. The mineralised zones are characterised by silica-ankerite/siderite-pyrite alteration, mainly in well foliated siliceous rocks. The mineralised zones are associated with pervasive silicification with local preservation of breccia textures that have been overprinted by the dominant S1 fabric. Higher gold grades appear to correlate well with the presence and abundance of pyrite, which appears to be spatially associated with the intersection of the NW trending D1 thrust surface, and a NE trending strain corridor.

The Pakaka mineralisation continues down plunge (18°) towards the NE beyond the limits of the drilling and represents further exploration potential. The Pakaka mineralisation extends over a strike length of 1,000 m, has a variable thickness and has been identified to a depth of 350 m below surface.

The ore can be interpreted as an open (gentle) fold with a thick (30 m) western limb dipping 7° toward SE and a very thin (average 12 m) eastern limb dipping 18° towards the SE (Figure 14-4). The axial plan is trending NE and plunging 18° towards the NE.

18^th September 2018 Page 151

Kibali Gold Mine NI 43-101 Technical Report

Pakaka mineralisation, like Mengu, is single mineralisation lode with internal high-grade zones.

Figure 14-4 3D View Pakaka Low-grade Mineralisation (Grey), High-grade Mineralisation (Orange) and

Optimised Pit Shells ($1,000 Dark Blue & $1,500 Light Blue)

Kombokolo

Kombokolo mineralisation is located between the clastic Meta-conglomerate unit and the footwall of the ironstone, following the brittle deformation within those units. Carbonate, sericite, silica, and chlorite alteration is associated with the mineralisation at varying intensities, plus disseminated sulphides, predominantly pyrites. High-grades generally sit in the very strong deformed meta-conglomerate with strong and dark chlorite alteration and high percentage of pyrite content. Grain size become very fine and foliations rarely identifiable.

Structurally, the mineralisation, in general, plunges low to moderate angles (30°) to the NE/ENE. The dip direction is mainly to the NW and dip angle is around 23° but flattens out in the up-plunge.

Nine low-grade lodes have been modelled, with the principal lodes being 1001 and 1002. These have a down plunge continuation of about 490 m within the $1,500 pit shell to a depth around 200 m below surface, an average width of 150 m, with an average thickness of 15 m which can reach about 25 m to 35 m locally. The lode remains open down plunge.

Two high-grade lodes have been identified in 2017 with geology and alteration criteria. They are 1101 and 1102 respectively within 1001 and 1002 now modelled as low-grade envelops. 1006 and 1007 have been intersected this year with three new diamond holes but they are below the current resources pit (Figure 14-5).

18^th September 2018 Page 152

Kibali Gold Mine NI 43-101 Technical Report

Figure 14-5 3D View of Kombokolo Mineralisation Within Both $1,000 & $1,500 Whittle Shells

18^th September 2018 Page 153

Kibali Gold Mine NI 43-101 Technical Report

Pamao

Pamao deposit has three main lodes hosted predominately within sandstone and sometimes in the BIF. Silica alteration dominants with ACSA alteration occurring locally with higher grade mineralisation, some chlorite is present. Pyrite is the dominant sulphide mineral.

Lithological sequence at Pamao include: A hanging wall basaltic unit with the lower contact marked in some places by carbonaceous shale (graphitic shear) interpreted as a thrust. The Meta-sandstone unit (middle formation) is in general strongly foliated with albite-silica-carbonate alteration with narrow layers of magnetite overprinting the early green schist facies. The magnetite alteration layer (potentially relic BIF) was not observed down dip as it was intersected in the diamond holes drilled up dip. This unit, host rock of the mineralised system, is intruded by a cent-metric porphyry unit strongly sheared. A coarse-grained material (sheared meta-conglomerate) with chlorotic matrix was identified as lower formation. Structurally, the ore bodies in general, plunge at low to moderate angles (25°) to the NE. The dip direction is mainly to the NNW and dip angle is generally very flat (5°).

The Pamao mineralisation model is presented in Figure 14-6.

Figure 14-6 3D View of Pamao Mineralisation

Mengu Hill

Mineralisation at Mengu is a single large domain with high-grade rod like shoots embedded within. Both wireframes were modelled separately and coded to represent the domain that they belong to. Mineralisation is predominantly hosted in the banded iron formation (BIF) which forms the cap of the hill (Figure 14-7). Lithological, weathering, redox and mineralisation wireframes are all built based on drill hole data and surface mapping.

The mineralisation wireframes were snapped to drilled holes as much as possible to create a more accurate interpretation. This resulted in minimising the influence of unwanted zones in the final composite files.

18^th September 2018 Page 154

Kibali Gold Mine NI 43-101 Technical Report

Figure 14-7 3D View of Mengu Mineralisation and Pit Shells

Mengu Village and Marakeke

Re-interpretations of the mineralisation at Mengu Village and Marakeke resulted in a single mineralised domain for each deposit.

The Mengu Village deposit mineralisation is approximately 150 m in strike length with an average thickness of 15 m and has been identified to a depth of 150 m below the surface. Re-interpretations of the mineralisation at Mengu Village resulted in a single mineralised domain for each deposit.

The Marakeke deposit occurs as a single tabular lens typically between 10 m to 30 m thick, that trends NW and dips gently to the NE. The mineralised zone tested by drilling has a strike length of approximately 1,000 m and extends 200 m down dip. Interpretations were carried out on north-south 80 m spaced sections depending on drilling spacing. At this grade cut-off and section spacing the resulting interpretation demonstrates consistent geometry and excellent continuity of the mineralised zone. Three 3D wireframes were generated defining the interpreted mineralised volume.

18^th September 2018 Page 155

Kibali Gold Mine NI 43-101 Technical Report

14.4 Topography

The Topography has been depicted using a 2 m digital terrain model (DTM) surface named ‘Kibali_lida_topo_combined_20101230.00t’ (clipped from the regional DTM completed by Kibali Goldmines in December 2010). This DTM covers the entire Project area as required for mine design purposes. The surface was checked visually against known drill hole collar elevations, and an acceptable match was found.

Original data was captured in UTM WGS84 Zone 35N with elevation. For the purposes of converting the elevation from UTM to Mine Grid, a scale factor of 5,000 m was applied to the elevation. Once the conversion was completed all data i.e. drill holes, DTM, 3D wireframes, and block models were checked to ensure that they all use the same mine grid system.

14.5 Bulk Density

Density (specific gravity) values were measured from diamond drill core samples by applying the Archimedean principles

density = weight (in air) ÷ (weight (in air) – weight (in water)

Densities are applied to the model via the lithological and weathering model. Whereby the density values within the selected lithology and weathering combination are averaged to give a density value to be applied into the block model via a script. The data is reviewed to remove any outliers that may exist and coded for the different mineralisation lodes. These outliers are noted to be mostly at the contacts of different weathering zones.

The depth of the weathering interfaces has been interpreted from drill holes and these are divided into four categories:

1 Fresh rock is the unweathered underlying lithology.

2 Transition materials are the first appearance of recognisable chips of the underlying lithology. These chips are clearly oxidised and constitutes mostly drill and blast material, depending on topography this transitional zone may be quite thin.

3 Oxide material is a zone of red/orange coloured silt/clay fragments with no recognisable lithology, generally clay rich.

4 Where there are no density measurements, or the volume of density data is not sufficient to make an unbiased estimate for the sub group, a substitute density is applied. This substitute density has been calculated using the density obtained from other lodes with similar rock and mineralisation characteristics.

These models are well represented with density data across all the weathering profiles.

Kibali performs quarterly truck factor tests and broken density calibration using weighbridge and cavity monitoring scanners (CMS).

18^th September 2018 Page 156

Kibali Gold Mine NI 43-101 Technical Report

KCD Open Pit

Table 14-19 summarises the assigned density values used for KCD open pit model in November 2017.

Table 14-19 KCD Open Pit Assigned Density Summary

Lodes	Weathering Code	Lithology		Min Density (g/cm3)	Max Density (g/cm3)	Mean Density (g/cm3)	No of Samples	Outliers Removed	Assigned Density (g/cm3)
3000	1 (FRESH)	Mcp	100	1.73	3.79	3.01	988	21	3.01
		CSS	300	2.19	3.49	2.89	112	8	2.89
		Chs	500	2.72	3.56	3.16	16	-	3.18
		Qch	550	2.59	3.63	3.09	173	1	3.09
		Sch	600	2.35	3.64	2.78	26	-	2.78
		Dol	800	2.73	3.17	2.91	8	54	2.90
	2 (TRANS)	Mcp	100	1.22	3.61	2.39	210	9	2.39
		CSS	300	1.94	2.27	2.13	9	3	2.26
		Chs	500	1.24	3.16	2.59	14	1	2.36
		Sch	600	1.53	2.62	2.09	4	2	2.13
		Dol	800	1.45	3.38	2.45	8	18	2.45
	3 (OXIDE)	Mcp	100	1.14	3.27	1.78	214	-	1.78
		Chs	500	1.14	2.76	1.90	30	-	2.01
5000	1 (FRESH)	Mcp	100	1.01	3.94	3.04	4,025	22	3.04
		MSI	250	2.82	2.82	2.82	1	2	2.83
		CSS	300	1.04	3.87	2.88	267	5	2.88
		Chs	500	1.69	3.90	3.18	1,085	6	3.19
		Qch	550	1.90	3.66	3.09	598	-	3.09
		Sch	600	2.71	3.57	2.98	8	2	2.78
		Dol	800	2.50	3.63	2.92	20	9	2.9
	2 (TRANS)	Mcp	100	1.54	3.19	2.41	59	6	2.41
		CSS	300	1.60	3.47	2.19	14	3	2.26
		Chs	500	1.62	2.19	1.96	5	3	2.36
		Qch	550	1.63	2.72	2.21	8	-	2.48
	3 (OXIDE)	Mcp	100	1.31	2.58	1.84	50	-	1.84
		CSS	300	1.87	1.87	1.87	1	-	1.89
		Chs	500	1.47	2.47	2.06	6	-	2.01
		Qch	550	1.77	1.77	1.77	1	-	1.80
9000	1 (FRESH)	Mcp	100	1.19	4.03	2.99	4,569	17	2.99
		CSS	300	1.75	3.72	2.83	340	1	2.83
		Chs	500	1.58	3.99	3.21	732	4	3.21
		Sch	600	2.72	3.41	2.86	22	-	2.77
		Dol	800	2.76	3.51	2.91	35	-	2.91
	2 (TRANS)	Mcp	100	1.72	2.71	2.30	5	1	2.3
		CSS	300	2.10	2.63	2.26	4	-	2.32
	3 (OXIDE)	Mcp	100	1.51	2.67	1.85	4	1	1.65
Waste	1 (FRESH)	Mcp	100	1.01	3.95	2.85	12,996	121	2.85
		MSI	250	2.58	3.22	2.83	210	-	2.83
		CSS	300	1.43	3.95	2.82	2,836	54	2.82
		Chs	500	1.64	3.98	3.20	1,231	2	3.20
		Qch	550	1.27	3.71	2.93	642	46	2.93
		Sch	600	1.53	3.32	2.75	199	4	2.75
		Dol	800	2.14	3.75	2.90	619	2	2.90
	2 (TRANS)	Mcp	100	1.08	3.58	2.10	757	192	2.10
		MSI	250	2.08	2.80	2.48	17	-	2.24
		CSS	300	1.05	3.27	2.01	253	118	2.01
		Chs	500	1.43	3.22	2.34	50	8	2.34
		Qch	550	1.34	3.24	2.24	35	13	2.24
		Sch	600	1.62	2.39	2.16	4	1	2.16
		Dol	800	1.21	3.01	2.13	38	13	2.13
	3 (OXIDE)	Mcp	100	1.03	3.91	1.61	979	30	1.61
		MSI	250	1.11	1.67	1.35	7	-	1.69
		CSS	300	1.23	2.73	1.90	22	2	1.89
		Chs	500	1.35	3.19	2.11	34	1	2.11
		Qch	550	1.16	2.42	1.81	6	-	1.81
		Dol	800	1.28	1.95	1.63	7	-	1.69
Total							34,613	806
** Updated in 2017

18^th September 2018 Page 157

Kibali Gold Mine NI 43-101 Technical Report

KCD Underground

Table 14-20 summarises the assigned density values used for KCD underground model in 2017.

Table 14-20 KCD Underground Assigned Density Summary

Lodes	Weathering Code	Lithology		Min Density (g/cm3)	Max Density (g/cm3)	Mean Density (g/cm3)	No of Samples	Outliers Removed	Assigned Density (g/cm3)
3000 Lodes	3	Mcp	100	1.14	2.59	1.72	220	18	1.72
		CSS	300	1.29	1.79	1.53	4	-	1.69**
		Chs	500	1.14	2.56	1.94	42	11	2.05**
		Chs	550	1.37	2.29	1.90	3	1	1.81**
	2	Mcp	100	1.80	3.00	2.39	129	54	2.32**
		CSS	300	1.87	2.73	2.24	30	9	2.32**
		Chs	550	2.39	2.91	2.69	9	3	2.48**
		Sch	600	1.94	1.94	2.62	227	3	2.32**
		Dol	800	1.81	1.81	2.70	235	4	2.46**
	1	Mcp	100	2.52	3.79	3.04	888	22	3.04
		CSS	300	2.52	3.34	2.88	85	7	2.88
		Chs	500	2.61	3.45	3.04	36	-	3.19**
		Chs	550	2.59	3.63	3.09	287	-	3.09**
		Sch	600	2.65	3.64	2.79	25	1	2.79
		Dol	800	2.85	3.31	2.98	8	-	2.91**
5000 lodes	3	Mcp	100	1.31	2.31	1.79	42	42	1.79
		CSS	300	1.77	1.77	1.77	1	1	1.69
		Chs	500	1.35	2.58	2.05	15	15	2.05
		Chs	550	1.16	2.05	1.51	3	3	1.81**
	2	Mcp	100	1.81	2.96	2.39	40	40	2.39
		CSS	300	1.87	2.71	2.15	14	14	2.32**
		Chs	500	1.84	2.96	2.21	13	13	2.48**
		Chs	550	1.95	2.94	2.42	7	7	2.48**
	1	Mcp	100	2.53	3.94	3.04	4,331	23	3.04
		CSS	300	2.66	3.87	2.94	267	267	2.94
		Chs	500	2.62	3.92	3.19	842	842	3.19
		Chs	550	2.57	3.66	3.09	672	6	3.09
		Sch	600	2.71	3.57	2.98	8	8	2.78**
		Dol	800	2.72	2.72	3.35	2.92	23	2.92
9000 lodes	3	Mcp	100	1.51	1.71	1.58	3	1	1.69**
	2	Mcp	100	2.02	2.02	2.71	2.36	5	2.32**
		CSS	300	2.10	2.10	2.63	2.37	2	2.32**
	1	Mcp	100	2.54	4.03	2.99	2,896	15	2.99
		CSS	300	2.56	3.38	2.83	155	-	2.83
		Chs	500	2.57	3.99	3.04	2,693	7	3.12**
		Sch	600	2.72	3.41	2.86	22	-	2.86
		Dol	800	2.71	3.37	2.88	41	-	2.88
Waste	3	Mcp	100	1.03	2.60	1.59	948	30	1.59
		CSS	300	1.19	2.31	1.70	44	2	1.70
		Chs	500	1.24	2.52	1.88	18	1	2.05**
		Chs	550	1.20	2.54	1.87	16	-	1.81**
		Dol	800	1.28	1.95	1.63	7	-	1.69*
	2	Mcp	100	1.80	3.00	2.30	525	194	2.30
		CSS	300	1.81	2.89	2.24	214	114	2.24
		Chs	500	1.87	2.99	2.38	31	9	2.44**
		Chs	550	1.83	2.96	2.45	50	13	2.45
		Sch	600	2.31	2.39	2.34	3	1	2.32**
		Dol	800	1.90	2.94	2.46	22	13	2.46
	1	Mcp	100	2.50	3.93	2.86	11,453	116	2.86
		MSI	250	3.14	3.14	3.14	1	-	2.83**
		CSS	300	2.50	3.95	2.83	2,358	54	2.83
		Chs	500	2.53	3.98	3.02	2,875	7	3.10**
		Chs	550	2.50	3.71	2.98	927	46	2.98
		Sch	600	2.52	3.32	2.76	195	4	2.76
		Dol	800	2.52	3.75	2.91	559	2	2.91
Total							34,648	2,075
** Updated in 2017

Sessenge

Table 14-21 summarises the assigned density values used at Sessenge.

18^th September 2018 Page 158

Kibali Gold Mine NI 43-101 Technical Report

Table 14-21 Sessenge Assigned Density Summary

Lodes	Weathering Code	Lithology	Min Density (g/cm3)	Max Density (g/cm3)	Mean Density (g/cm3)	No of Samples	Outliers Removed	Assigned Density (g/cm3)
Ore	1	100	2.56	3.76	2.99	218	-	2.99
	1	500	2.66	3.54	3.11	43	-	3.11
	1	600	2.61	2.84	2.73	14	-	2.86
	2	100	1.85	2.95	2.52	23	-	2.32
	2	500	2.36	3.00	2.69	5	-	2.48
	3	100	1.52	2.60	2.14	4	-	1.69
	3	500	2.43	2.43	2.43	1	-	2.05
Waste	1	100	2.61	3.76	2.88	381	2	2.88
	1	500	2.56	3.53	3.14	77	-	3.14
	2	100	1.85	2.87	2.48	12	3	2.48
	2	500	1.98	3.00	2.61	15	1	2.61
	3	100	1.52	2.52	2.19	6	1	1.69
	3	500	2.43	2.60	2.51	3	8	2.05
Total						802	15

** Updated in 2017

Gorumbwa

Table 14-22 summarises the assigned density values used at Gorumbwa.

Table 14-22 Gorumbwa Assigned Density Summary

Lodes	Weathering	Lithology	Min Density (g/cm3)	Max Density (g/cm3)	Mean Density (g/cm3)	No of Samples	Outliers Removed	Assigned Density (g/cm3)
Ore	1	MCP(100)	2.68	3.23	2.82	99	1	2.82
	1	MCP_Fe(120)	2.74	3.18	2.85	11	-	2.81**
	1	MCP_He(150)	2.81	2.81	2.81	1	-	2.81**
	1	MSS(200)	2.64	19.02	2.85	1,899	26	2.85
	1	Sch(600)	2.73	2.95	2.79	15	-	2.79
	1	Dol(800)	2.66	2.99	2.83	39	-	2.83
	2	MSS(200)	1.81	2.73	2.15	5	3	2.26**
	3	MCP(100)	1.24	1.76	1.50	2	-	1.53**
	3	MSS(200)	1.21	1.66	1.46	5	-	1.57**
Waste	1	MCP(100)	2.61	13.29	2.81	3,574	31	2.81
	1	MCP_Fe(120)	2.62	3.25	2.81	555	2	2.81
	1	MCP_He(150)	2.69	3.64	2.81	643	8	2.81
	1	MSS(200)	2.60	4.69	2.81	7,394	76	2.81
	1	Sch(600)	2.68	3.06	2.78	331	7	2.78
	1	Dol(800)	2.64	3.17	2.83	571	5	2.83
	2	MCP(100)	1.80	2.80	2.40	57	59	2.40
	2	MCP_Fe(120)	2.10	2.78	2.35	3	1	2.39**
	2	MCP_He(150)	1.80	2.80	2.39	31	10	2.39
	2	MSS(200)	1.80	2.78	2.26	63	51	2.26
	2	Sch(600)	2.12	2.78	2.64	7	4	2.26**
	2	Dol(800)	2.26	2.77	2.52	2	2	2.26**
	3	MCP(100)	1.01	2.38	1.53	183	17	1.53
	3	MCP_He(150)	1.22	2.27	1.62	17	3	1.62
	3	MSS(200)	1.08	2.35	1.57	111	6	1.57
	3	Sch(600)	1.27	1.99	1.51	9	2	1.57**
Total						15,627	314

** Updated in 2017

18^th September 2018 Page 159

Kibali Gold Mine NI 43-101 Technical Report

Pakaka

Table 14-23 summarises the assigned density values used at Pakaka.

Table 14-23 Pakaka Assigned Density Summary

Lodes	Weathering	Lithology	Min Density (g/cm3)	Max Density (g/cm3)	Mean Density (g/cm3)	No of Samples	Outliers Removed	Assigned Density (g/cm3)
Ore	1	200	2.48	3.53	2.81	275	1	2.81
	1	300	2.86	2.86	2.86	1	-	2.81
	1	500	2.80	3.48	3.09	4	-	2.91
	2	200	1.66	2.79	2.50	12	1	2.37
	3	200	1.06	2.62	1.72	14	-	1.63
Waste	1	200	2.72	3.09	2.89	15	-	2.81
	1	200	2.44	4.01	2.83	1,104	6	2.83
	1	300	2.49	3.25	2.78	101	3	2.78
	1	500	2.65	3.57	2.91	107	-	2.91
	1	550	2.73	3.22	2.97	2	-	2.91
	1	700	2.50	3.22	2.81	455	6	2.81
	2	200	1.66	2.95	2.37	34	5	2.37
	2	300	1.63	2.85	2.55	10	1	2.37
	2	500	1.63	3.23	2.07	7	6	2.37
	2	700	1.61	3.60	2.29	84	18	2.29
	3	200	1.06	2.75	1.63	83	2	1.63
	3	300	1.45	1.60	1.52	3	-	1.58
	3	700	1.04	2.78	1.58	199	5	1.58
Total						2,510	54

** Updated in 2017

Kombokolo

Table 14-24 summarises the assigned density values used at Kombokolo.

Table 14-24 Kombokolo Assigned Density Summary

Lodes	Weathering	Lithology	Min Density (g/cm3)	Max Density (g/cm3)	Mean Density (g/cm3)	No of Samples	Outliers Removed	Assigned Density (g/cm3)
Ore	1	100	2.59	13.12	2.97	332	16	2.97
	1	500	2.51	4.49	3.12	31	5	3.12
	1	600	2.71	3.51	2.98	10	2	2.81
	2	100	2.17	2.44	2.28	5	1	2.30**
	2	500	2.53	2.79	2.66	2	1	2.59**
Waste	1	100	2.50	14.40	2.82	1,044	41	2.82
	1	250	2.50	9.56	2.74	236	32	2.74
	1	300	2.50	3.71	2.75	254	14	2.75
	1	500	2.61	6.25	3.23	94	4	3.23
	1	600	2.67	3.64	2.81	35	1	2.81
	1	800	2.62	3.03	2.85	43	4	2.85
	2	100	2.15	2.33	2.25	5	30	2.4**
	2	250	1.80	3.00	2.44	58	39	2.44
	2	300	2.09	2.58	2.43	5	-	2.30
	2	500	1.98	2.45	2.18	4	1	2.44**
	2	800	2.16	2.78	2.50	3	-	2.30**
	3	100	1.45	2.02	1.63	21	-	1.63
	3	250	1.01	2.60	1.64	192	11	1.64
Total						2,374	202

** Updated in 2017

Pamao

Table 14-25 summarises the assigned density values used at Pamao.

18^th September 2018 Page 160

Kibali Gold Mine NI 43-101 Technical Report

Table 14-25 Pamao Assigned Density Summary

Lodes	Weathering	Lithology	Min Density (g/cm3)	Max Density (g/cm3)	Mean Density (g/cm3)	No of Samples	Outliers Removed	Assigned Density (g/cm3)
Ore	1	200	2.70	3.16	2.86	32	-	2.86
	1	500	3.15	3.29	3.24	3	-	3.05
	2	200	2.57	2.99	2.80	31	8	2.37**
	2	500	2.90	2.90	2.90	1	4	2.37**
	3	200	2.35	2.35	2.35	1	8	1.63**
	1	200	2.51	59.61	3.08	274	4	3.08
Waste	1	300	2.66	2.83	2.73	21	1	2.73
	1	500	2.68	3.15	3.01	15	-	3.05
	1	700	2.67	2.87	2.78	24	1	2.78
	2	200	2.29	2.96	2.77	90	1	2.37**
	2	300	1.93	2.81	2.60	11	-	2.37**
	2	700	2.35	2.79	2.61	3	-	2.29**
	3	200	1.02	2.23	1.81	3	37	1.63**
	3	300	2.31	2.31	2.31	1	-	1.63**
	1	200	2.70	3.16	2.86	32	-	2.86**
Total						510	64

** Updated in 2017

Mengu Hill

Table 14-26 summarises the assigned density values used at Mengu Hill.

Table 14-26 Mengu Assigned Density Summary

Lodes	Weathering	Lithology	Min Density (g/cm3)	Max Density (g/cm3)	Mean Density (g/cm3)	No of Samples	Outliers Removed	Assigned Density (g/cm3)
Ore	1	100	2.68	3.57	3.10	64	-	3.10
	1	250	2.69	3.86	3.10	47	-	3.10
	1	500	2.74	3.93	3.29	372	3	3.29
	1	600	2.54	3.59	2.94	63	-	2.94
	2	100	1.95	2.86	2.47	6	3	2.26**
	2	250	2.52	2.92	2.72	2	3	2.35**
	2	500	2.21	2.99	2.54	65	147	2.54
	3	100	1.07	2.14	1.60	26	1	1.60
	3	250	2.07	2.51	2.29	2	1	1.68**
	3	500	1.17	2.53	1.77	35	6	1.77
Waste	1	100	2.51	3.91	2.88	802	16	2.88
	1	250	2.52	3.22	2.79	597	36	2.79
	1	500	2.66	3.92	3.28	881	7	3.28
	1	600	2.54	3.91	2.88	430	14	2.88
	2	100	1.80	2.89	2.25	109	92	2.25
	2	250	1.80	2.86	2.34	41	45	2.34
	2	500	2.20	2.99	2.53	84	181	2.53
	2	600	1.83	2.87	2.56	20	24	2.56
	3	100	1.11	2.53	1.50	107	3	1.50
	3	250	1.07	2.55	1.41	99	-	1.50**
	3	500	1.08	2.48	1.69	10	-	1.75**
	3	600	1.09	2.53	1.78	15	-	1.68**
Total						3,877	582

** Updated in 2017

Mengu Village and Marakeke

At Mengu Village, the following assumed bulk densities were applied:

☐ Oxide – 1.65 t/m³

☐ Transitional – 2.3 t/m³

☐ Fresh – 2.8 t/m³

18^th September 2018 Page 161

Kibali Gold Mine NI 43-101 Technical Report

At Marakeke, the following assumed bulk densities were applied:

☐ Oxide – 1.65 g/cm³

☐ Transitional – 2.3 g/cm³

☐ Fresh – 2.8 g/cm³

14.6 Compositing

Prior to selecting the composite length, the data was visually analysed using a histogram of sample length to identify the mode of length. The coefficient of variation, standard deviation, and mean plots were produced with several composite lengths to ensure that they remain stable and do not increase with compositing.

A 2.0 m length composite was applied to all drill holes that intersected the mineralisation wireframes at Kibali. A minimum composite length of 0.5 m was used so that any residual samples below this length were disregarded during resource estimation. Compositing is completed in Maptek Vulcan software using the merge option for small composites, which add the last composite, if it is small enough, to the previous interval. A tolerance length is defined, and anything less than the specified tolerance length will be added to the previous interval in the same geology. For Kibali, a tolerance length of 0.5 m is used.

18^th September 2018 Page 162

Kibali Gold Mine NI 43-101 Technical Report

KCD

Figure 14-8 illustrates a log histogram and log probability plot of the gold grades for the 2 m uncapped composites within all mineralised domains at KCD.

Figure 14-8 KCD Log Histogram and Log Probability Plot of 2 m Uncapped Composites Within Mineralised Lodes

Sessenge

Figure 14-9 illustrates a log histogram and log probability plot of the gold grades for the 2 m uncapped composites within all mineralised domains at Sessenge.

Figure 14-9 Sessenge Log Histogram and Log Probability Plot of 2 m Uncapped Composites Within Mineralised Lodes

18^th September 2018 Page 163

Kibali Gold Mine NI 43-101 Technical Report

Gorumbwa

Figure 14-10 illustrates a log histogram and log probability plot of the gold grades for the 2 m uncapped composites within all mineralised domains at Gorumbwa.

Figure 14-10 Gorumbwa Log Histogram and Log Probability Plot of 2 m Uncapped Composites Within Mineralised Lodes

Pakaka

Figure 14-11 illustrates a log histogram and log probability plot of the gold grades for the 2 m uncapped composites within all mineralised domains at Pakaka.

Figure 14-11 Pakaka Log Histogram and Log Probability Plot of 2 m Uncapped Composites Within Mineralised Lodes

18^th September 2018 Page 164

Kibali Gold Mine NI 43-101 Technical Report

Kombokolo

Figure 14-12 illustrates a log histogram and log probability plot of the gold grades for the 2 m uncapped composites within all mineralised domains at Kombokolo.

Figure 14-12 Kombokolo Log Histogram and Log Probability Plot of 2 m Uncapped Composites Within Mineralised Lodes

Pamao

Figure 14-13 illustrates a log histogram and log probability plot of the gold grades for the 2 m uncapped composites within all mineralised domains at Pamao.

Figure 14-13 Pamao Log Histogram and Log Probability Plot of 2 m Uncapped Composites Within Mineralised Lode

Mengu Hill

Figure 14-14 illustrates a log histogram and log probability plot of the gold grades for the 2 m uncapped composites within all mineralised domains at Mengu.

18^th September 2018 Page 165

Kibali Gold Mine NI 43-101 Technical Report

Figure 14-14 Mengu Log Histogram and Log Probability Plot of 2 m Uncapped Composites Within Mineralised Lodes

Mengu Village and Marakeke

For both Mengu Village and Marakeke 2.5 m downhole composites were utilised.

18^th September 2018 Page 166

Kibali Gold Mine NI 43-101 Technical Report

14.7 Treatment of High-grades (Top Cutting)

Top cutting was applied to reduce the effect of high-grade outliers during resource estimation. Generally, the top cutting occurred within the top percentile ranges, between the 95^th to 99.9^th percentiles within the individual mineralised lodes. A multi-variate analysis method was used to select the top cap analysing a combination of histograms, probability plot, and disintegration.

KCD

At KCD, a total of 152,065 samples were included in the database for top cutting analysis. In total, 442 samples were top cut between 3.29 g/t Au and 100.00 g/t Au. Top cutting reduced the average mean grade from 2.90 g/t Au to 2.79 g/t Au and resulted in a reduction of the coefficient of variation from 2.90 to 1.98. In total, the metal reduction was -4% overall. A detailed breakdown of the statistical analysis for top cutting at KCD is presented in Table 14-27.

Table 14-27 KCD Top Cutting Analysis

Domain	No of   Samples	Minimum   Au (g/t)	Maximum   Au (g/t)	Mean   Au (g/t)	CV Au   (g/t)	Capped   Au (g/t)	Mean   Au Cap   (g/t)	CV Au Cap   (g/t)	No of   Samples   Capped	Metal   Reduction
3001	4751	0.005	112.04	0.73	3.54	9	0.63	1.79	30	-14%
3002	14,745	0.005	132.88	1.55	2.39	40	1.51	1.97	18	-3%
3003	20,113	0.005	499.70	1.79	3.38	79	1.75	2.67	15	-2%
3004	4,675	0.005	33.00	2.02	1.76	27	2.02	1.72	9	0%
3005	521	0.005	22.06	1.04	1.67	3.5	0.88	0.93	21	-15%
3101	2,287	0.005	102.18	3.83	1.73	50	3.76	1.60	10	-2%
3102	4,521	0.005	180.00	5.4	1.61	80	5.33	1.46	9	-1%
3103	74	0.248	16.070	2.96	0.83	9	2.85	0.74	1	-4%
3105	277	0.005	75.70	3.48	2.23	30	3.08	1.69	6	-11%
3106	800	0.010	514.37	10.27	2.27	40	8.68	1.23	47	-15%
3107	2,214	0.030	329.87	10.00	1.99	100	9.50	1.58	15	-5%
3108	695	0.020	25.48	4.20	0.90	20	4.17	0.88	6	-1%
3109	142	0.080	89.80	5.72	1.51	22	5.20	0.92	2	-9%
3110	98	0.030	56.00	4.32	1.78	10	3.02	1.01	10	-30%
3111	22	0.150	14.02	3.47	1.04	14.02	3.47	1.01	0	0%
3112	63	0.600	58.81	6.19	1.74	58.81	6.19	1.69	0	0%
5002	1743	0.005	67.27	1.00	2.29	17.8	0.97	1.77	2	-3%
5003	7,356	0.005	240.00	2.46	2.86	70	2.37	2.28	17	-4%
5004	966	0.005	120.83	0.61	7.63	10	0.36	3.64	12	-41%
5005	27,726	0.005	433.59	1.35	3.51	70	1.31	2.44	8	-3%
5006	705	0.005	40.69	0.93	2.80	8.2	0.79	1.75	9	-15%
5007	1101	0.005	25.90	1.21	1.72	20	1.20	1.65	2	-1%
5101	11,950	0.005	3,008.00	7.94	3.75	100	7.57	1.18	19	-5%
5102	4,965	0.006	172.47	6.25	1.30	70	6.19	1.21	8	-1%
5104	336	0.022	240.00	9.32	2.38	70	8.11	1.55	5	-13%
5105	6,361	0.005	727.02	5.71	2.78	76	5.38	1.35	11	-6%
5106	56	0.401	14.68	3.30	0.84	8	3.08	0.68	3	-7%
5107	521	0.009	89.3.	3.45	1.85	22	3.22	1.09	3	-7%
5110	1,314	0.009	540.00	6.82	2.74	49	6.09	1.35	15	-11%
9002	69	0.020	17.02	2.56	1.30	17.02	2.56	1.35	0	0%
9003	48	0.010	6.48	1.34	1.21	6.48	1.34	1.19	0	0%
9004	23,067	0.005	90.09	0.77	2.72	15.00	0.73	2.14	69	-5%
9008	15	0.18	3.29	1.61	0.65	3.29	1.61	0.65	0	0%
9101	2,063	0.013	100.46	6.67	1.3	50.00	6.60	1.24	9	-1%
9102	468	0.03	65.58	4.69	1.25	24.00	4.50	1.03	6	-4%
9103	113	0.01	40.7	4.26	1.25	40.70	4.26	1.24	0	0%
9105	5,124	0.005	200.68	5.67	1.41	47.00	5.49	1.16	45	-3%
Total	152,065	0.005	3,008	2.90	2.83	100.00	2.79	1.98	442	-4%

18^th September 2018 Page 167

Kibali Gold Mine NI 43-101 Technical Report

Sessenge

At Mengu, a total of 7,807 samples were included in the database for top cutting analysis. In total, 21 samples were top cut between 6.50 g/t Au and 25.00 g/t Au. Top cutting reduced the average mean grade from 2.11 g/t Au to 2.09 g/t Au and resulted in a reduction of the coefficient of variation from 1.29 to 1.24. The total metal reduction was negligible. A detailed breakdown of the statistical analysis for top cutting at Sessenge is presented in Table 14-28.

Table 14-28 Sessenge Top Cutting Analysis

Domain	No of   Samples	Min Au   (g/t)	Max Au   (g/t)	Mean   Au (g/t)	CV Au   (g/t)	Capped   Au (g/t)	Mean   Au (g/t)	CV Au   Cap	No of   Samples   Capped	% Metal   Reduction
9001	126	0.05	9.98	2.21	0.81	6.50	2.15	0.74	4	-3%
9002	4,704	0.02	45.00	1.54	1.46	25.00	1.53	1.38	3	-1%
9003	309	0.005	18.81	1.51	1.37	9.00	1.43	1.10	4	-5%
9007	419	0.005	5.71	1.19	0.83	5.71	1.19	0.83	0	0%
9008	478	0.005	26.00	1.95	1.64	20.00	1.92	1.57	4	-2%
9009	272	0.02	19.6	1.33	1.50	9.00	1.25	1.20	4	-6%
9102	771	0.03	31.00	3.93	0.81	19.00	3.90	0.78	2	-1%
9103	728	0.05	21.30	4.98	0.63	21.30	4.98	0.63	0	0%
9001	126	0.05	9.98	2.21	0.81	6.50	2.15	0.74	4	-1%
Total	7,807	0.005	45	2.11	1.29	25	2.09	1.24	21	-3%

Gorumbwa

At Gorumbwa, a total of 7,331 samples were included in the database for top cutting analysis. In total, 73 samples were top cut between 8.00 g/t Au and 33.00 g/t Au. Top cutting reduced the average mean grade from 2.71 g/t Au to 2.47 g/t Au and resulted in a reduction of the coefficient of variation from 2.57 to 1.69. A detailed breakdown of the statistical analysis for top cutting at Gorumbwa can be found in Table 14-29.

Table 14-29 Gorumbwa Top Cutting Analysis

Domain	No of   Samples	Min Au   (g/t)	Max Au   (g/t)	Mean Au   (g/t)	CV Au   (g/t)	Capped   Au (g/t)	Mean Cap   Au (g/t)	CV   Cap Au	No of   Samples   Capped	% Metal   Reduction
1001	1,512	0.005	36.20	2.4	1.55	22	2.36	1.43	13	-2%
1002	419	0.019	12.70	1.37	1.06	12.7	1.37	1.04	0	0%
1003	549	0.005	240.00	2.15	4.87	17	1.7	1.46	2	-21%
1004	2,757	0.005	174.98	3.49	2.47	33	3.13	1.7	33	-10%
1005	93	0.03	46.40	2.65	2.42	14	2.03	1.33	2	-23%
1006	759	0.025	94.20	2.22	2.49	21.00	1.98	1.67	7	-11%
1007	215	0.005	91.30	5.01	2.20	26.00	4.14	1.61	7	-17%
1008	636	0.03	22.70	1.80	1.43	22.7	1.80	1.41	0	0%
1009	107	0.05	16.35	1.56	1.57	8.00	1.42	1.23	2	-9%
1010	72	0.04	27.60	3.41	1.61	15.00	2.98	1.46	4	-13%
1011	155	0.015	4.34	0.70	0.97	4.34	0.70	0.95	0	0%
1012	57	0.22	45.90	4.20	1.95	15.00	3.30	1.30	3	-21%
Total	7,331	0.01	240	2.71	2.57	33	2.47	1.69	73	-9%

18^th September 2018 Page 168

Kibali Gold Mine NI 43-101 Technical Report

Pakaka

At Pakaka, a total of 21,357 samples were included in the database for top cutting analysis. In total, 34 samples were top cut between 10.00 g/t Au and 34.00 g/t Au. Top cutting reduced the average mean grade from 2.37 g/t Au to 2.31 g/t Au and resulted in a reduction of the coefficient of variation from 2.19 to 1.47. In total, the metal reduction was -3%, overall. A detailed breakdown of the statistical analysis for top cutting at Pakaka can be found in Table 14-30.

Table 14-30 Pakaka Top Cutting Analysis

Domain	No of   Samples	Min Au   (g/t)	Max Au   (g/t)	Mean Au   (g/t)	CV Au   (g/t)	Capped   Au (g/t)	Mean Au   (g/t)	CV Au   Cap	No of   Samples   Capped	% Metal   Reduction
1001	17,140	0.01	65.42	1.50	1.44	32.00	1.49	1.36	13	-1%
1007	163	0.03	6.99	1.23	1.00	6.99	1.23	0.98	0	0%
1101	2,613	0.01	89.43	7.02	0.96	34.00	6.93	0.87	11	-1%
1102	101	0.37	520.00	10.4	5.06	13.00	4.36	0.74	8	-58%
1103	199	0.79	24.60	3.97	0.84	10.00	3.70	0.62	9	-7%
1105	1,141	0.01	60.00	4.13	1.20	21.00	3.89	0.82	19	-6%
Total	21,357	0.01	520.00	2.37	2.19	34.00	2.31	1.47	60	-3%

Kombokolo

At Kombokolo, a total of 6,364 samples were included in the database for top cutting analysis. In total, eight samples were top cut between 13.4 g/t Au and 39.60 g/t Au. Top cutting reduced the average mean grade from 3.03 g/t Au to 2.89 g/t Au and resulted in a reduction of the coefficient of variation from 1.75 to 1.46. A detailed breakdown of the statistical analysis for top cutting at Kombokolo can be found in Table 14-31.

Table 14-31 Kombokolo Top Cutting Analysis

Domain	No of   Samples	Minimum   Au (g/t)	Maximum   Au (g/t)	Mean Au   (g/t)	CV Au   (g/t)	Capped   Au (g/t)	Mean Au   (g/t)	CV Au   Cap	No of   Samples   Capped	% Metal   Reduction
1001	2,824	0.005	72.2	1.46	1.94	17.1	1.39	1.41	12	-5%
1002	1,518	0.01	34.7	1.75	1.55	13.4	1.68	1.29	14	-4%
1003	57	0.02	5.76	1.26	0.93	5.76	1.26	0.92	0	0%
1004	15	0.34	3.43	0.97	0.84	3.43	0.97	0.82	0	0%
1005	33	0.005	9.88	2.61	1.13	9.88	2.61	1.13	0	0%
1006	5	0.50	3.91	1.80	0.77	3.91	1.80	0.78	0	0%
1007	2	1.18	2.63	2.08	0.54	2.63	2.08	0.49	0	0%
1101	1,247	0.009	79.70	6.82	1.14	39.6	6.65	1.00	9	-2%
1102	663	0.057	124.4	5.66	1.32	18.7	5.15	0.80	18	-9%
Total	6,364	0.005	124.4	3.03	1.75	39.6	2.89	1.46	53	-5%

18^th September 2018 Page 169

Kibali Gold Mine NI 43-101 Technical Report

Pamao

At Pamao, a total of 2,182 samples were included in the database for top cutting analysis. No top cut was applied as there were no identified outliers within the distribution. A detailed breakdown of the statistical analysis for top cutting at Pamao can be found in Table 14-32.

Table 14-32 Pamao Top Cutting Analysis

Domain	No of   Samples	Minimum   Au (g/t)	Maximum   Au (g/t)	Mean Au   (g/t)	CV Au   (g/t)	Capped   Au (g/t)	Mean   Au   Cap	CV Au   Cap	No of   Samples   Capped	% Metal   Reduction
2001	186	0.01	3.67	0.80	0.80	3.67	0.80	0.79	0	0%
2002	1,348	0.01	26.8	1.40	1.41	26.8	1.40	1.41	0	0%
2003	648	0.01	21.4	1.26	1.4	21.4	1.26	1.39	0	0%
Total	2,182	0.01	26.8	1.30	1.41	26.80	1.30	1.40	0	0%

Mengu Hill

At Mengu, a total of 17,343 samples were included in the database for top cutting analysis. In total, 10 samples were top cut between 4.40 g/t Au and 65.00 g/t Au. Top cutting reduced the average mean grade from 3.20 g/t Au to 2.19 g/t Au and resulted in a reduction of the coefficient of variation from 1.64 to 2.58. The total metal reduction was negligible. A detailed breakdown of the statistical analysis for top cutting at Mengu is presented in Table 14-33.

Table 14-33 Mengu Top Cutting Analysis

Domain	No of   Samples	Min   Au   (g/t)	Max   Au   (g/t)	Mean   Au   (g/t)	CV   Au   (g/t)	Capped   Au (g/t)	Mean   Au   (g/t)	CV Au   Cap	No of   Samples   Capped	% Metal   Reduction
1001	9,467	0.01	33.60	1.14	1.43	30.00	1.14	1.42	1	0%
1002	125	0.03	4.40	1.19	0.95	4.40	1.19	0.91	0	0%
1101	7,587	0.02	117.00	5.76	1.19	65.00	5.74	1.14	5	0%
1102	25	1.00	31.10	6.69	1.10	10.00	4.89	0.58	4	-27%
1103	139	0.05	11.30	3.16	0.69	11.30	3.16	0.67	0	0%
Total	17,343	0.01	117.00	3.20	1.64	65.00	3.19	1.58	10	0%

Mengu Village and Marakeke

For Mengu Village, a high-grade assay cut of 10 g/t Au was applied to the 2.5 m downhole composites representing a point above the 99^th percentile of the single mineralised gold population. Only two composites out of 256 were affected by the cut, which resulted in a reduction in the total metal content of approximately 1%.

Grade top cutting at Marakeke is set at 15 g/t Au, although the procedure for determining these in not available. In the QP’s opinion, the grade top cutting level at Marakeke is reasonable when compared to other deposit at Kibali.

18^th September 2018 Page 170

Kibali Gold Mine NI 43-101 Technical Report

14.8 Variography

Exploratory Data Analysis (EDA) was conducted using Snowden Supervisor statistical software and all modelling and estimation was completed in Maptek Vulcan. The drill data is stored in an industry standard Maxwell DataShed SQL database connected with Microsoft Access via ODBC link. Values less than the detection limit (<0.01 g/t Au) were replaced with 0.005 g/t Au.

Variography has been used to analyse the spatial continuity and relation within the individual mineralised lodes and to determine the appropriate search strategy and estimation parameters. The Variogram modelling process involved the following steps:

☐ A normal score transform was applied to all data prior to undertaking variography on the top capped, declustered composite dataset; The data was transformed into a normal score space using Snowden Supervisor.

☐ Calculate and model the omni-directional or down hole variogram to characterise the nugget effect;

☐ Systematically calculate orientated variogram in three dimensions to identify the plane of greatest continuity;

☐ Calculate a variogram fan within the plane of greatest continuity to identify the direction of maximum continuity within this plane.

☐ Model experimental variogram in the direction of maximum continuity and the orthogonal directions.

☐ Apply a back transform to all variogram models to obtain the appropriate variogram models for interpolation of raw composite data.

Within the domains, the relative nuggets ranged between 10% and 40% indicating a low to moderate grade variability, which is typical for these type of gold deposits. Variogram ranges interpreted were typically significantly greater than the average drill hole spacing.

In some areas which contain infill grade control drilling, such as KCD, variograms were required for nested structures thus multiple ranges were used.

Where an individual domain has insufficient samples to undertake variography, the variography parameters from a comparative domain with a similar trend was used and the orientation adjusted to match the domain with insufficient data.

KCD

At KCD, the three lodes return significantly different variography results. The lodes are broadly categorised as the upper 3000 lodes, 5000 lodes, and at the deeper 9000 lodes.

3000 Lodes

At the KCD 3000 lodes, the relative nugget effect ranged from 24% to 26% for the high-grade domains, and from 15% to 26% for the low grade domains. Figure 14-15 illustrates an example of the KCD 3102 normal score domain and Figure 14-16 presents the nested back transformed variogram models.

18^th September 2018 Page 171

Kibali Gold Mine NI 43-101 Technical Report

Figure 14-15 KCD 3102 Normal Score Variogram Models

Figure 14-16 KCD 3102 Nested Back Transformed Variogram Model

5000 Lodes

At the KCD 5000 lodes, the relative nugget effect ranged from 8% to 14% for the high-grade domains, and from 5% to 22% for the low grade domains. Figure 14-17 illustrates an example of

18^th September 2018 Page 172

Kibali Gold Mine NI 43-101 Technical Report

the KCD 5101 normal score domain and Figure 14-18 presents the nested back transformed variogram models.

Figure 14-17 KCD 5101 Normal Score Variogram Models

Figure 14-18 KCD 5101 Nested Back Transformed Variogram Model

18^th September 2018 Page 173

Kibali Gold Mine NI 43-101 Technical Report

9000 Lodes

At the KCD 9000 lodes, the relative nugget effect ranged from 11% to 13% for the high-grade domains, and was 7% for the low grade domains. Figure 14-19 illustrates an example of the KCD 9105 normal score domain and Figure 14-20 presents the nested back transformed variogram models.

Figure 14-19 KCD 9105 Nested Back Transformed Variogram Model

Figure 14-20 KCD 9105 Nested Back Transformed Variogram Model

18^th September 2018 Page 174

Kibali Gold Mine NI 43-101 Technical Report

Sessenge

At Sessenge, the relative nugget effect ranged was 22% for both high-grade and low-grade domains. Figure 14-21 illustrates an example of the Sessenge 9002 normal score domain and Figure 14-22 presents the nested back transformed variogram models.

Figure 14-21 Sessenge 9002 Normal Score Variogram Models

Figure 14-22 Sessenge 9002 Nested Back Transformed Variogram Model

18^th September 2018 Page 175

Kibali Gold Mine NI 43-101 Technical Report

Gorumbwa

At Gorumbwa, the relative nugget effect ranged from 11% to 13%. Figure 14-23 illustrates an example of the Gorumbwa 1001 normal score domain and Figure 14-24 presents the nested back transformed variogram models.

Figure 14-23 Gorumbwa 1001 Normal Score Variogram Models

Figure 14-24 Gorumbwa 1001 Nested Back Transformed Variogram Model

18^th September 2018 Page 176

Kibali Gold Mine NI 43-101 Technical Report

Pakaka

At Pakaka, the relative nugget effect ranged from 16% to 20%. Figure 14-25 illustrates an example of the Pakaka 1001 normal score domain and Figure 14-26 presents the nested back transformed variogram models.

Figure 14-25 Pakaka 1001 Normal Score Variogram Models

Figure 14-26 Pakaka 1001 Nested Back Transformed Variogram Model

18^th September 2018 Page 177

Kibali Gold Mine NI 43-101 Technical Report

Kombokolo

At Kombokolo, the relative nugget effect ranged from 36% to 44%. Kombokolo is observed to have a higher nugget effect than the other domains at Kibali. Figure 14-27 illustrates an example of the Kombokolo 1101 and 1002 normal score domain and Figure 14-28 presents the nested back transformed variogram models.

Figure 14-27 Kombokolo 1101 and 1002 Normal Score Variogram Models

Figure 14-28 Kombokolo 1101 and 1002 Nested Back Transformed Variogram Model

18^th September 2018 Page 178

Kibali Gold Mine NI 43-101 Technical Report

Pamao

At Pamao, the relative nugget effect was 28%. Figure 14-29 illustrates an example of the Pamao normal score domain and Figure 14-30 presents the nested back transformed variogram models. The down hole variogram model was applied to direction 3.

Figure 14-29 Pamao Normal Score Variogram Models

Figure 14-30 Pamao Nested Back Transformed Variogram Model

18^th September 2018 Page 179

Kibali Gold Mine NI 43-101 Technical Report

Mengu Hill

At Mengu Hill, the relative nugget effect ranged from 8% to 20% Figure 14-31 illustrates an example of the Mengu Hill 1001 normal score domain and Figure 14-32 presents the nested back transformed variogram models.

Figure 14-31 Mengu Hill 1001 Normal Score Variogram Models

Figure 14-32 Mengu Hill 1001 Nested Back Transformed Variogram Model

18^th September 2018 Page 180

Kibali Gold Mine NI 43-101 Technical Report

Variogram Validation

Prior to interpolation runs, each semi-variogram model is cross validated to ensure that any bias in estimated grades compared to the actual sample grades are minimal. This was checked by estimating a grade value at each composite sample point, which ignored said sample point. The resulting grade is compared to the actual sample grade in the same location and is plotted on a scatter plot to establish a possible trend or bias and relative standard error. In most cases, there is level of smoothing in an estimated grade compared to the actual sample grade, but overall, estimated grades and sample grades match well and conditional bias is minimal.

14.9 Block Model Estimation

QKNA

Ordinary Kriging (OK) was used to estimate the resources. QKNA was applied to help to determine the optimal block size, minimum number of samples, search radius and, block discretisation for each domain. Figure 14-33 illustrates the results of the QKNA for domain 9103 at Sessenge.

Figure 14-33 QKNA for Sessenge Domain 9103 Open Pit GC Zone

18^th September 2018 Page 181

Kibali Gold Mine NI 43-101 Technical Report

KCD

Due to the large number of resource, grade control and advance grade control domains present at KCD, a small subset of the KCD QKNA parameters are detailed in Table 14-40.

Table 14-34 QKNA Parameters for KCD 5003 Domain

Domain	Block Size (m)			Run	Search Radius (m)			No. Samples		Discretisation			High- Grade Restriction (g/t)	HY Restriction
	X	Y	Z		Y	X	Z	Min	Max	X	Y	Z		X	Y	Z
5003 GC	5	5	2.5	1	20	15	5	12	22	2	2	2
				2	47	32	10	8	16				17.00	17	15	11
				3	94	64	15	4	12				17.00	17	15	11
				4	150	100	23	2	10				17.00	17	15	11
5003 AdvGC	10	10	5	1	28	16	12	12	22	4	4	2	17.00	17	15	11
				2	47	32	10	8	16				17.00	17	15	11
				3	94	64	15	4	12				17.00	17	15	11
				4	150	100	23	2	10				17.00	17	15	11
5003 Res	10	20	10	1	28	16	12	12	22	4	4	2
				2	47	32	10	8	16
				3	94	64	15	4	12
				4	150	100	23	2	10

Sessenge

A small subset of the Sessenge QKNA parameters are detailed in Table 14-35.

Table 14-35 QKNA Parameters for Sessenge 9002 Domain

Domain	Block Size (m)			Run	Search Radius (m)			No. Samples		Discretisation			High- Grade Restriction (g/t)	HY Restriction
	X	Y	Z		Y	X	Z	Min	Max	X	Y	Z		X	Y	Z
9002 GC	5	5	2.5	1	26	15	5	8	18	2	2	2	14	20	10	5
				2	45	18	8	6	15				14	20	10	5
				3	68	25	7	4	12				14	20	10	5
				4	98	35	11	1	8				14	20	10	5
9002 AdvGC	5	10	5	1 2 3 4	35	15	5	8	18	2	4	2	14	20	10	5
					45	18	8	6	15				14	20	10	5
					68	25	7	4	12				14	20	10	5
					98	35	11	1	8				14	20	10	5
9002 Res	10	20	5	1	70	30	8	8	20	4	8	2	14	20	10	5
				2	70	30	8	6	18				14	20	10	5
				3	95	50	12	4	14				14	20	10	5
				4	150	75	12	2	12				14	20	10	5

18^th September 2018 Page 182

Kibali Gold Mine NI 43-101 Technical Report

Gorumbwa

A small subset of the Gorumbwa QKNA parameters are detailed in Table 14-36. No high-grade restrictions were used at Gorumbwa.

Table 14-36 QKNA Parameters for Gorumbwa 1001 Domain

Domain	Block Size (m)			Run	Search Radius (m)			No. Samples		Discretisation
	X	Y	Z		Y	X	Z	Min	Max	X	Y	Z
1001 GC	5	5	2.5	1	32	28	6	8	20	5	5	2
				2	64	56	18	6	14
				3	80	80	20	4	12
1001 Res	10	10	5	1	40	40	10	8	26	5	5	4
				2	110	48	40	8	18
				3	220	96	80	6	14

Pakaka

A small subset of the Pakaka QKNA parameters are detailed in Table 14-37. No high-grade restrictions were used at Pakaka.

Table 14-37 QKNA Parameters for Pakaka 1001 Domain

Domain	Block Size (m)			Run	Search Radius (m)			No. Samples		Discretisation
	X	Y	Z		Y	X	Z	Min	Max	X	Y	Z
1001 GC	5	5	2.5	1	33	18	12	10	20	5	5	2
				2	49	27	18	8	18
				3	66	36	24	6	16
				4	99	54	36	6	16
1001 AdvGC	10	20	10	1	33	18	12	10	20	4	5	4
				2	49	27	18	8	18
				3	66	36	24	6	16
				4	98	38	14	6	16
1001 Res	10	20	10	1	49	27	18	8	18	4	5	4
				2	66	36	24	6	16
				3	99	54	36	4	16
				4	159	91	25	4	16

18^th September 2018 Page 183

Kibali Gold Mine NI 43-101 Technical Report

Kombokolo

A small subset of the Kombokolo QKNA parameters are detailed in Table 14-38.

Table 14-38 QKNA Parameters for Kombokolo 1001 Domain

Domain	Block Size (m)			Run	Search Radius (m)			No. Samples		Discretisation			High-Grade Restriction (g/t)	HY Restriction
	X	Y	Z		Y	X	Z	Min	Max	X	Y	Z		X	Y	Z
1001 GC	5	5	2.5	1	15	10	5	8	18	3	3	3
				2	30	20	10	6	16				11.1	15	10	5
				3	60	40	20	4	14				11.1	15	10	5
				4	120	80	40	4	12				11.1	15	10	5
				5	240	160	80	4	10				11.1	15	10	5
				6	480	320	160	4	8				11.1	15	10	5
1001 AdvGC	5	10	2.5	1	30	20	10	8	22	3	3	3
				2	60	40	20	6	20				11.1	30	20	10
				3	120	80	40	4	18				11.1	30	20	10
				4	240	160	80	4	16				11.1	30	20	10
				5	480	320	160	4	14				11.1	30	20	10
				6	960	640	320	4	12				11.1	30	20	10
1001 Res	10	15	5	1	30	20	10	8	26	3	3	3
				2	60	40	20	6	24				11.1	30	20	10
				3	120	80	40	4	22				11.1	30	20	10
				4	240	160	80	4	20				11.1	30	20	10
				5	480	320	160	4	18				11.1	30	20	10
				6	960	640	320	4	16				11.1	30	20	10

Pamao

The full set of Pamao QKNA parameters are detailed in Table 14-36. No high-grade restrictions were used at Pamao.

Table 14-39 QKNA Parameters for Gorumbwa 2001 Domain

Domain	GC / Res	Block Size (m)			Run	Search Radius (m)			No. Samples		Discretisation
		X	Y	Z		Y	X	Z	Min	Max	X	Y	Z
2001 Res	RES	20	20	5	1	30	15	10	4	18	3	3	3
					2	60	30	20	4	18
					3	120	60	40	2	18
					4	240	120	40	2	18
					5	480	240	40	2	18
2002 Res	RES	20	20	5	1	30	15	10	6	20	3	3	3
					2	60	30	20	6	20
					3	120	60	40	4	20
					4	240	120	40	4	20
					5	480	240	40	2	20
2003 Res	RES	20	20	5	1	30	15	10	4	18	3	3	3
					2	60	30	20	4	18
					3	120	60	40	2	18
					4	480	240	40	2	18
					5	240	120	40	2	18

Historical Resources

Mengu Village and Marakeke Mineral Resources were estimated using the Uniform Conditioning (UC) methodology with allowance for an Information Effect incorporating important modifying factors such as likely grade control drilling, mining selectivity, and cut-off grade criteria. The application of UC technique is based on the premise that mining would be by open pit extraction.

18^th September 2018 Page 184

Kibali Gold Mine NI 43-101 Technical Report

A Selective Mining Unit (SMU) of 5 m by 5 m by 2.5 m was evaluated within OK panels 20 m by 20 m by 5 m for the purposes of reporting local recoverable open pit resources. An Information Effect has been applied to the SMU blocks with the assumption of grade control drilling on a 7.5 m by 7.5 m by 2.5 m sampling grid.

The UC estimation was completed by Cube Consulting Pty Ltd (Cube) and all the documentation regarding the UC work can be found in the Technical Report (NI 43-101), 20th May 2010, Adams et al., complied by Cube. No drilling has been completed on these deposits since acquisition by Kibali Goldmines in 2009.

UC, as implemented in Isatis V7 software, was applied to these estimates with an appropriate change of support to incorporate an SMU of 5 m by 5 m by 2.5 m (high) and an information effect adjustment to take into account a likely grade control programme to enable reporting of Mineral Resources above a range of grade cut-offs.

The cut-off range used in each deposit was 0.0, 0.25, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0, 1.1, 1.2, 1.3, 1.4, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5, 5.0, 10.0, and 15.0 g/t gold. If insufficient data were available within a domain, parameters from the most similar domain in the deposit area were adopted.

The UC process requires the calculation of a number of support correction coefficients to determine the likely distribution of mining SMU of a specified dimension within larger estimation grade panels. Cube has employed a standardised methodology for all domains to determine the required parameters including average panel variance, SMU variance, and covariance.

To determine a robust average panel variance required for the panel support correction of each domain, Cube calculates statistics on the OK estimated panels. This variable describes the variability of panel grades within a domain. The OK estimated panel grades are generally more smoothed than the expected real panel grades due to the smoothing effects of kriging. The amount of smoothing is influenced by a number of factors including the variogram, the block dimensions, and the data configuration. Ideally, a panel support correction would be calculated for each panel to take into account its specific data configuration. The theoretical mean variance of well estimated panels (defined by a slope of regression greater than 0.75) will provide a robust measure of panel variance estimated with a typical data configuration.

14.10 Block Models

Setup

Consideration is given for selectivity during mine design and planning when selecting an appropriate block size considering the geology and domaining used.

The purpose of the sub-blocking is to better define the geological and domained contacts within the block model, by allowing a higher resolution when the model is interpolated. The models were not regularised back to the nominal 10 m by 20 m by 10 m before reporting resources.

18^th September 2018 Page 185

Kibali Gold Mine NI 43-101 Technical Report

The search strategy used was based on the variogram results obtained through considering the data distribution for each of the domains. The search ellipsoids were orientated optimally for each domain, considering the plunge and dip of the wireframe.

Each pass is completed using a varying degree of restrictions before any given block can be estimated. In total, four passes were used on every block model, each with increasing search radius representing the decreasing confidence in the blocks for each subsequent run.

Dykes were wireframed and coded into both the block with the relevant grade field (au_ok) set to zero as default.

Table 14-40 tabulates the block model variables and attributes that were coded to the block model either prior to, or during each interpolation run for OK models.

Table 14-40 Block Model Variables and Attributes

Variables	Default	Type	Description
domain	9999	integer	Domain or zone of mineralisation 9999 = data coverage, 3001 - 9108 =ore zones
au_ok	0	double	Estimated grade from Kriging
density	0	double	Density
sr	0	double	Slope of regression
ns	0	integer	Number of sample used in estimation
kv	0	double	Kriging variance
bv	0	float	Block variance = average sample variance - block sample variance
ke	0	float	Kriging efficiency
lm	0	float	Lagrange multiplier
dist_ans	0	float	Anisotropic distance to nearest sample
dist_cat	0	float	Cartesian distance to closest sample
no_hole	0	integer	Number of holes in estimation
no_pass	0	integer	Number of estimation run
wt_sum	0	float	Sum of weights
wt_mean	0	float	Weight of the mean
oxidation	1	integer	Oxidation 1= fresh 2= transition 3= oxide
depletion	0	integer	Depletion on model 1 = depleted 0 = Insitu
lith	999	integer	Litho unit 100 = mcp, 200 = msi,300 = css, 500 = chs, 600 = sch, 800 = dol, 999 = undefined
redox	2	integer	Redox front 1= oxidised 2 = reduced
au_orig	0	float	Original estimated grade before depletion
mined_out	0	integer	Mined zone yymmdd = year and month e.g. (121001)
lodes	9999	integer	Mineralisation lodes
gd_ans	0	float	Grade of the closest sample (anisotropic)
gd_cat	0	float	Grade of the closest sample (Cartesian)
nw_sum	0	float	Sum of negative weights
pw_sum	0	float	Sum of positive weights
kw_min	0	float	Minimum Kriging weight
kw_max	0	float	Maximum Kriging weight
avd_ans	0	float	Average anisotropic distance to sample derived from search ellipsoid
wvd_ans	0	float	Weighted average anisotropic distance to sample derived from search ellipsoid
avd_cat	0	float	Cartesian average distance to samples
wvd_cat	0	float	Weighted Cartesian average distance to samples
rescat	0	integer	Classification 1 = Measured 2 = Indicated 3 = Inferred 4 = exploration target 0 = undefined
depletion	0	integer	Depletion from old working on model 1 = depleted 0 = Insitu

KCD

The resource block model for KCD is named ‘kcd_res_2017_09_25’. The block model has a parent block size of 10 m by 20 m by 10 m with a minimum sub cell size of 1.25 m by 1.25 m by 1.25 m. This takes into account that most of the higher grade open pit drill holes were on a 10 m

18^th September 2018 Page 186

Kibali Gold Mine NI 43-101 Technical Report

by 5 m grid spacing. Underground drilling was drilled on an approximate 15 m by 20 m spacing. The block model was flagged by each mineralisation domain separately by priority. Table 14-41 summarises the KCD block model extents.

Table 14-41 KCD Global Block Model Extent (With Rotation)

Block Extents	Easting(X)	Northing(Y)	Elevation(Z)
Origin	784,008	343,540	5,000
Minimum Offset	0	0	0
Maximum Offset	2,120	4,400	1,100
Parent Block Size (m)	10	20	10
Sub Cell Size (m)	1.25	1.25	1.25
Rotation (Degrees)	135	0	0

In total, KCD contains 101 estimation domains. The search ellipsoid was orientated individually for each estimation domain. An example of the search ellipsoid orientation for KCD is presented in Table 14-42.

Table 14-42 KCD Domain 1001 and 1002 Search Ellipsoid Orientation

Domain	Estimation Domains	Domain Orientation
		Bearing (°)	Plunge (°)	Dip (°)
5101	510112	43	-22	56
	510121	35	-41	58
	510122	51	-11	53
	510124	47	-21	38
	510125	43	-23	33
	510126	45	-20	40
	510127	36	-35	54
5102	510211	49	-15	60
	510221	44	-15	60
	510241	39	-25	45
	510242	42	-19	25

Sessenge

The resource block model for Sessenge is named ‘ses_res_2017_08_25’. The block model has a parent block size of 10 m by 10 m by 5 m with a minimum sub cell size of 1.25 m by 1.25 m by 1.25 m. The block model was flagged by each mineralisation domain separately by priority. Table 14-43 summarises the Sessenge block model extents.

Table 14-43 Sessenge Global Block Model Extent (No Rotation)

Block Extents	Easting(X)	Northing(Y)	Elevation(Z)
Origin	784,700	343,460	5,500
Minimum Offset	0	0	0
Maximum Offset	1,250	1,740	500
Parent Block Size (m)	10	20	5
Sub Cell Size (m)	1.25	1.25	1.25
Rotation (Degrees)	135	0	0

The Sessenge search ellipsoid was orientated individually for each estimation domain (Table 14-44).

18^th September 2018 Page 187

Kibali Gold Mine NI 43-101 Technical Report

Table 14-44 Sessenge Search Ellipsoid Orientation

Domain	Estimation Domains	Domain Orientation
		Bearing (°)	Plunge (°)	Dip (°)
9001	900131	45	-25	10
9002	900211	45	-25	17
	900221	45	-25	17
	900222	43	-20	-14
	900223	43	-20	-5
	900231	54	-18	-15
	900232	55	-19	5
9003	900321	44	-22	-2
	900322	42	-16	-24
	900331	48	-21	4
	900332	52	-17	-16
9009	900921	38	-25	-11
9102	910211	41	-21	15
	910221	41	-21	15
9103	910311	50	-28	11
	910321	50	-28	11

Gorumbwa

The resource block model for Gorumbwa is named ‘gor_res_2017_06_30’. The block model has a parent block size of 10 m by 10 m by 5 m with a minimum sub cell size of 1.25 m by 1.25 m by 1.25 m. The block model was flagged by each mineralisation domain separately by priority. Table 14-45 summarises the Gorumbwa block model extents.

Table 14-45 Gorumbwa Global Block Model Extent (No Rotation)

Block Extents	Easting(X)	Northing(Y)	Elevation(Z)
Origin	785,000	344,400	5,200
Minimum Offset	0	0	0
Maximum Offset	1,000	1,500	900
Parent Block Size (m)	10	10	5
Sub Cell Size (m)	1.25	1.25	1.25
Rotation (Degrees)	135	0	0

The Gorumbwa search ellipsoid was orientated individually for each estimation domain (Table 14-46).

18^th September 2018 Page 188

Kibali Gold Mine NI 43-101 Technical Report

Table 14-46 Gorumbwa Search Ellipsoid Orientation

Domain	Domain Orientation
	Bearing (°)	Plunge (°)	Dip (°)
1001	49	-20	29
1002	48	-18	31
1003	48	-18	25
1004	48	-23	29
1005	46	-5	22
1006	35	-20	30
1007	42	-22	38
1008	45	-29	7
1009	47	-23	20
1010	45	-10	27
1011	45	-10	12
1012	45	-7	18

Pakaka

The resource block model for Pakaka is named ‘pak_res_2017_09_15’. The block model has a parent block size of 10 m by 20 m by 10 m with a minimum sub cell size of 1.25 m by 1.25 m by 1.25 m. The block model was flagged by each mineralisation domain separately by priority. Table 14-47 summarises the Pakaka block model extents.

Table 14-47 Pakaka Global Block Model Extent (No Rotation)

Block Extents	Easting(X)	Northing(Y)	Elevation(Z)
Origin	787,880	347,800	5,400
Minimum Offset	0	0	0
Maximum Offset	1,400	1,500	700
Parent Block Size (m)	20	10	5
Sub Cell Size (m)	2.5	2.5	0.5
Rotation (Degrees)	90	0	0

The Pakaka search ellipsoid was orientated individually for each estimation domain (Table 14-48).

Table 14-48 Pakaka Search Ellipsoid Orientation

Domain	Domain Orientation
	Bearing (°)	Plunge (°)	Dip (°)
1001	42	-17	-12
1007	44	-17	28
1101	42	-17	-11
1102	38	0	11
1103	40	-10	-20
1104	39	-11	-21
1105	43	-17	0

Kombokolo

The resource block model for Kombokolo is named ‘kom_res_2017_12_23’. The block model has a parent block size of 10 m by 15 m by 5 m with a minimum sub cell size of 1.25 m by 1.25 m by 0.625 m. The block model was flagged by each mineralisation domain separately by priority. Table 14-49 summarises the Kombokolo block model extents.

18^th September 2018 Page 189

Kibali Gold Mine NI 43-101 Technical Report

Table 14-49 Kombokolo Global Block Model Extent (No Rotation)

Block Extents	Easting(X)	Northing(Y)	Elevation(Z)
Origin	786,000	345,000	5,500
Minimum Offset	0	0	0
Maximum Offset	1,100	1,125	900
Parent Block Size (m)	10	15	5
Sub Cell Size (m)	1.25	1.25	0.625
Rotation (Degrees)	90	0	0

The Kombokolo search ellipsoid was orientated individually for each estimation domain (Table 14-50).

Table 14-50 Kombokolo Search Ellipsoid Orientation

Domain	Domain Orientation
	Bearing (°)	Plunge (°)	Dip (°)
1101	66	-25	30
1102	66	-25	30
1001	66	-30	25
1002	55	-30	25
1003	48	-27	25
1004	48	-27	25
1005	45	-24	23
1006	60	-25	25
1007	60	-25	25

Pamao

The resource block model for Pamao is named ‘Pam_res_2017_07_04’. The block model has a parent block size of 20 m by 20 m by 5 m with a minimum sub cell size of 2.5 m by 2.5 m by 0.5 m. The block model was flagged by each mineralisation domain separately by priority listed. Table 14-51 summarises the Pamao block model extents.

Table 14-51 Pamao Global Block Model Extent (No Rotation)

Block Extents	Easting(X)	Northing(Y)	Elevation(Z)
Origin	786,600	348,400	5,400
Minimum Offset	0	0	0
Maximum Offset	1,800	1,000	600
Parent Block Size (m)	20	20	5
Sub Cell Size (m)	2.5	2.5	0.5
Rotation (Degrees)	90	0	0

The Pamao search ellipsoid was orientated individually for each estimation domain (Table 14-52).

Table 14-52 Pamao Search Ellipsoid Orientation

Domain	Domain Orientation
	Bearing (°)	Plunge (°)	Dip (°)
2001	20	-20	-15
2002	22	-20	-35
2003	20	-20	-30

Mengu Hill

The resource block model for Mengu is named ‘mgh_res_2016_10_24’. The block model has a parent block size of 10 m by 10 m by 5 m with a minimum sub cell size of 2.5 m by 2.5 m by

18^th September 2018 Page 190

Kibali Gold Mine NI 43-101 Technical Report

0.625 m. The block model was flagged by each mineralisation domain separately by priority. Table 14-53 summarises the Mengu block model extents.

Table 14-53 Mengu Global Block Model Extent (No Rotation)

Block Extents	Easting(X)	Northing(Y)	Elevation(Z)
Origin	782,510	350,610	5,300
Minimum Offset	0	0	0
Maximum Offset	1,100	1,300	700
Parent Block Size (m)	10	10	5
Sub Cell Size (m)	2.5	2.5	0.625
Rotation (Degrees)	0	0	0

The Mengu Hill search ellipsoid was orientated individually for each estimation domain (Table 14-54).

Table 14-54 Mengu Hill Search Ellipsoid Orientation

Domain	Domain Orientation
	Bearing (°)	Plunge (°)	Dip (°)
2001	35	-20	-25
2002	35	-20	-35
2003	35	-20	-35

Mengu Village and Marakeke

At Mengu Village and Marakeke the block model size was 20 m by 20 m by 5 m. The Mengu Village and Marakeke deposits were estimated using UC methodology with an allowance for an Information Effect, incorporating important modifying factors such as likely grade control drilling, mining selectivity, and cut-off grade criteria. Due to Mengu Village and Marakeke deposits representing 0.4% and 0.7% respectively of the combined Kibali resource, any change in the tonnages or grades through different estimation methods will have no material impact on the total resource.

The application of UC technique is based on the premise that mining would be by open pit extraction. An SMU of 5 m by 5 m by 2.5 m (X, Y, Z) was evaluated within OK panels of 20 m by 20 m by 5 m for the purposes of reporting local recoverable open pit resources. An Information Effect has been applied to the SMU blocks with the assumption of grade control drilling on a 7.5 m by 7.5 m by 2.5 m sampling grid. The UC estimation was completed by Cube and all the documentation regarding the UC work can be found in by Adams et al, 2010.

14.11 Resource Classification

Current Resources

Under the CIM definitions (CIM Standards on Mineral Resources and Reserves Definitions and Guidelines, 2014), and the 2012 Australian Code of Mineral Resources and Ore Reserves (2012 JORC Code) Measured Resources require that “quantity, grade, density, shape, and physical characteristics need to be established with confidence sufficient to allow the appropriate application of technical and economic parameters” such that production planning and the evaluation of the economic viability of the deposit is possible. In the case of Indicated Resources,

18^th September 2018 Page 191

Kibali Gold Mine NI 43-101 Technical Report

the level of confidence should be sufficient to allow for the application of appropriate technical and economic parameters, mine planning, and economic evaluation. Resource Classification was based on geological continuity and data density as well as estimation quality in form of slope of regression (SR) and kriging efficiency (KE). This was carried out by displaying the estimated blocks (SR and KE) together with the supporting data as guide.

The Mineral Resources are classified as Measured, Indicated, and Inferred Mineral Resources based on drilling density, geological continuity and confidence, the variogram range continuity and the slope of regression. The classification parameters are presented in Table 14-55.

Table 14-55 Kibali Resource Classification Parameters

Statistic	Deposit	Measured	Indicated	Inferred
Minimum Samples		8	6	4
Minimum Consecutive Sections		4	Good Geological Continuity	-
Maximum Drilling Density	KCD OP (m)	10 by 5 or 20 by 5	40 by 30	80 by 80
	KCD UG (m)	20 by 10	40 by 40	80 by 80
	Pakaka (m)	20 by 10 or 20 by 5	40 by 40	80 by 60
	Sessenge (m)	-	30 by 20	80 by 80
	Pamao (m)	-	20 by 40	80 by 80
	Gorumbwa (m)	10 by 5 or 15 by 10	20 by 10 or 30 by 30	80 by 80
	Kombokolo (m)	10 by 5 or 10 by 10	30 by 30	80 by 80

For Indicated Mineral Resources there are some allowances for areas where drilling density is lower but successive drilling campaigns have shown there is grade and geological continuity.

Mengu Village and Marakeke

Mengu Village was classified as Indicated by 40 m by 50 m drill spacing with Inferred at 80 m by 80 m or 120 m by 100 m drill spacing

Marakeke was classified as Inferred based on 80 m by 80 m drill spacing.

14.12 Block Model Depletion

Active mining areas are scanned using cavity monitoring laser scanners on a monthly basis and detailed drone photometry surface scans are completed on a weekly basis.

Every block model was flagged with the regional 2 m DTM ‘Kibali_lida_topo_ combined_20101230.00t’ with any blocks falling above the surface being flagged as air.

KCD

As an active mine, KCD requires depletion to represent the blocks mined until the end of the reporting period. Depletion pit surveys at KCD were updated in December 2017 and used to flag the block model in the mined_out field. The KCD underground resource block model is not depleted due to the sub blocking size required to accurately define the depletion. To deplete the model, all stopes and development void solids would require sub blocking into the model resulting

18^th September 2018 Page 192

Kibali Gold Mine NI 43-101 Technical Report

in a very large model size. For depletion, the Sarbanes-Oxley (SOX) depletion is removed from the measured material. Further work is underway to find an improved approach to depleting KCD underground model.

Sessenge

No depletion was applied to the model due to no historical or current mining at Sessenge.

Gorumbwa

Gorumbwa has been depleted using a sonar survey conducted in 2016 and the results of the 2016 drilling campaign that targeted mined voids. This new depletion has provided a higher degree of confidence in the mined-out shape. Areas scanned include the mine shaft which is also incorporated into the final Geotech model. Additional void scanning is planned to be completed during 2018.

Pakaka and Kombokolo

Pakaka and Kombokolo are both currently in operation and are depleted with pit surfaces as of 31^st December 2017.

Mengu Hill

Mining of Mengu Hill was complete in Q2 2017 and the model was depleted with the end of pit surface.

Mengu Village and Marakeke

No depletion was applied to the model due no historical or current mining at Mengu Village or Marakeke.

14.13 Block Model Validation

Once a block model had been classified, the following procedures were undertaken to check the block models and estimated grades in order to indicate any major errors during the estimation process, as well as testing the precision, accuracy, and any bias of the estimated grade:

a) A volume comparison between the block model estimation domains and related wireframes is undertaken. Table 14-56 summarises the variances between the wireframe and block model volumes across all deposits.

18^th September 2018 Page 193

Kibali Gold Mine NI 43-101 Technical Report

Table 14-56 Block Model Volume Comparison 2017

Domain	Wireframe Volume (m3)	Block Model  Volume (m3)	Variances
KCD	124,845,154	125,768,430	-1.15%
Pakaka	11,275,549	11,284,252	0.0%
Sessenge	9,046,054	9,051,912	0.1%
Kombokolo	2,560,128	2,559,498	0.0%
Gorumbwa	5,605,771	5,594,320	0.0%
Pamao	9,659,334	9,657,044	0.0%

b) A check of the number of the blocks estimated using the negative kriging weight is completed. Any blocks estimated using negative kriging weight have been reset to the anisotropic nearest block grade of the closest sample.

c) A comparison between the data minimum, maximum, mean, declustered mean and the estimation mean for each of the domains (within the open pit or underground reporting areas) is created. This is completed to check for possible over or under estimation.

d) Swath plots are created for each geological domain to validate the estimated grade variability compared to the composite along X, Y and Z axis. This is to check that the model estimate follows the trends seen in the data and that there is no general bias with over or under estimation. Areas with less data support are also highlighted for further drilling and geological work. The swath plots for Kibali show the confidence for the deposit is within acceptable limits and that conditional bias is kept to a minimum. An example for KCD along the X axis is shown in Figure 14-34.

e) A comprehensive visual check is undertaken comparing the data to the block estimates to check for an acceptable correlation, including checking that local high-grade trends on the block estimated is supported by data.

Figure 14-34 KCD SWATH Plot of Domains 5102 Along Y Axis

18^th September 2018 Page 194

Kibali Gold Mine NI 43-101 Technical Report

14.14 Resource Cut-Off Grades

KCD Open Pit Resources

The cut-off grade calculations for the KCD open pit Mineral Resources are broken down in Table 14-57.

Table 14-57 KCD 2017 Optimisation Parameters

Material Type	Unit	Oxide	Trans	Fresh	Total
Waste Cost	$/t mined	2.92	2.97	3.09	2.99
Extra Ore Cost -GC+Ore-Rehandle + Overhaul	$/t mined	1.27	1.27	1.27	1.27
GC Only	$/t mined	0.75	0.75	0.75	0.75
Dilution	%	10%	10%	10%	10%
Ore Loss	%	3%	3%	3%	3%
Haulage Cost	$/t mined	0.00	0.00	0.00	0.00
Process Cost	$/t milled	14.34	14.34	16.10	14.93
Processing Recovery	%	90.1	90.1	86.1	88.8
Plant Throughput	Mtpa
General/Admin	$/t milled	7.40	7.40	7.40	7.40
Gold Price (Resource)	$/oz	1,500	1,500	1,500	1,500
Total Process Cost	$/t	14.34	14.34	16.10	14.93
Total Mining Cost	$/t	16.15	16.40	17.01	16.52
Marginal In-situ Cut-off Grade	g/t Au	0.60	0.60	0.68	0.63
Strip Ratio					4.1
Full Grade Ore In-situ Cut-off Grade	g/t Au	0.99	1.00	1.11	1.03

KCD Underground Resources

The cut-off grade calculations for KCD underground Mineral Resources are broken down in Table 14-58.

Table 14-58 KCD Underground 2017 Optimisation Parameters

Material Type	Unit	Fresh
Mine Production	$/t mined	41.0
Capital	$/t mined	0.00
Backfill	$/t mined	0.00
Process Cost	$/t milled	16.1
Processing Recovery	%	85%
General/Admin	$/t milled	7.4
Gold Royalties	$/oz	52.5
Gold Price (Resource)	$/oz	1500
Total Unit Cash Cost	$/t	64.5
Mining Cut Off Grade	g/t Au	1.6

18^th September 2018 Page 195

Kibali Gold Mine NI 43-101 Technical Report

The cut-off grade for KCD underground remains at 1.6 g/t Au for 2017 resource reporting. There are some changes from 2016 which are summarised below:

☐ Mine Production changed from $44.10/t to $41.00/t.

☐ Recovery decreased from 89% to 85%.

☐ G&A cost increased from $6.80/t to $7.40/t.

☐ Gold Royalties increased from $35/oz to $52.5/oz.

☐ Total Unit Cash Cost reduced from $67.00/t to $64.50/t.

KCD Underground Optimised Minable Stope Shapes

The KCD 2017 underground Mineral Resources were calculated within optimised minable stope shapes at a cut-off grade of 1.6 g/t Au, and within the underground reporting box wireframe solid, with varying RL. This varying RL now limits the 5000 lode, 5680 mRL, and 3000 lode reporting to 5682.5 mRL for the 2017 Mineral Resource. This varying RL has been put in place to ensure that all material that forms part of the underground Mineral Resource is excluded from the Open pit Mineral Resource.

A reconciliation between the Mineral Resource reported using a block model cut-off grade reporting method against the applied MSO constraint is shown in the Table 14-59.

Figure 14-35 illustrates visual checks undertaken on blocks that were not included in the MSO shapes primarily due to geology and the shapes of mineralisation lodes. These blocks would have been included in the Mineral Resource estimation if a cut-off grade approach had been used.

18^th September 2018 Page 196

Kibali Gold Mine NI 43-101 Technical Report

Table 14-59 KCD Underground 2017 Optimisation Parameters

Class	2017 BM GT Res Model (kcd_res_2017_09_25) Depleted to EOY 2017			2017 MSO Constrained Res Model   (kcd_res_2017_09_25) Depleted to EOY 2017			Variances			Material Difference
	Tonnes	Grade (Au g/t)	Au Ounces	Tonnes	Grade (Au g/t)	Au Ounces	Tonnes	Grade (Au g/t)	Au Ounces	Tonnes	Grade (Au g/t)	Au Ounces
Measured	11,502,240	6.03	2,229,494	11,933,390	5.57	2,135,978	0.04	-8%	-4%	431,150	(6.75)	(93,517)
Indicated	51,151,578	4.97	8,171,986	65,262,331	3.64	7,638,032	28%	-27%	-7%	14,110,754	(1.18)	(533,953)
M+I	62,653,817	5.16	10,401,480	77,195,721	3.94	9,774,010	23%	-24%	-6%	14,541,904	(1.34)	(627,470)
Inferred	24,086,416	3.26	2,521,381	22,117,917	2.83	2,015,014	-8%	-13%	-20%	(1,968,499)	8.00	(506,367)

18^th September 2018 Page 197

Kibali Gold Mine NI 43-101 Technical Report

Figure 14-35 KCD 3D View of MSO Shapes Against Underground Blocks Above 1.6 g/t Au – View Towards SE

18^th September 2018 Page 198

Kibali Gold Mine NI 43-101 Technical Report

Figure 14-36 highlights the thin areas of mineralisation containing blocks above 1.6 g/t Au in 3000 which are excluded from MSO resource because they do not reach the cut-off grade within a mineable shape.

Figure 14-36 KCD 3D View of MSO Shapes Against Grade Blocks – View Towards SE

18^th September 2018 Page 199

Kibali Gold Mine NI 43-101 Technical Report

Figure 14-37 highlights the isolated blocks of high-grade in 5000, which are excluded from MSO resource because they do not reach the cut-off grade within a mineable shape due to the dyke and the inclusion of waste on the deposit boundary.

Figure 14-37 KCD 3D View of MSO Shapes Against Grade Blocks – View Towards SE

18^th September 2018 Page 200

Kibali Gold Mine NI 43-101 Technical Report

Figure 14-38 highlights the isolated blocks of high-grade in 9000, which are excluded from MSO resource because they do not reach the cut-off grade within a mineable shape.

Figure 14-38 KCD 3D Vew of MSO Shapes Against Grade Blocks – View Towards SE

18^th September 2018 Page 201

Kibali Gold Mine NI 43-101 Technical Report

Sessenge

The cut-off grade calculations for Sessenge open pit Mineral Resource are broken down in Table 14-60.

Table 14-60 Sessenge 2017 Optimisation Parameters

Material Type	Unit	Oxide	Trans	Fresh	Total
Waste Cost	$/t mined	2.62	2.68	2.80	2.70
Extra Ore Cost -GC+Ore-Rehandle + Overhaul	$/t mined	1.24	1.24	1.24	1.24
GC Only	$/t mined	0.75	0.75	0.75	0.75
Dilution	%	10%	10%	10%	10%
Ore Loss	%	3%	3%	3%	3%
Haulage Cost	$/t mined	0.00	0.00	0.00	0.00
Process Cost	$/t milled	14.34	14.34	16.10	14.93
Processing Recovery	%	90.3	75.9	79.1	81.8
Plant Throughput	Mtpa
General/Admin	$/t milled	7.40	7.40	7.40	7.40
Gold Price (Resource)	$/oz	1,500	1,500	1,500	1,500
Total Process Cost (per ore tonne mined)	$/t ore	14.34	14.34	16.10	14.93
Total Mining Cost (per ore tonne mined)	$/t ore	11.19	11.42	11.87	11.50
Marginal In-situ Cut-off Grade	g/t Au	0.60	0.72	0.74	0.68
Strip Ratio					2.8
Full Grade Ore In-situ Cut-off Grade	g/t Au	0.86	1.03	1.06	0.98

Gorumbwa

The cut-off grade calculations for Gorumbwa open pit Mineral Resource are broken down in Table 14-61.

Table 14-61 Gorumbwa 2017 Optimisation Parameters

Material Type	Unit	Oxide	Trans	Fresh	Total
Waste Cost	$/t mined	2.92	3.14	3.24	3.10
Extra Ore Cost -GC+Ore-Rehandle + Overhaul	$/t mined	1.28	1.28	1.28	1.28
GC Only	$/t mined	0.75	0.75	0.75	0.75
Dilution	%	10%	10%	10%	10%
Ore Loss	%	3%	3%	3%	3%
Haulage Cost	$/t mined	0.00	0.00	0.00	0.00
Process Cost	$/t milled	14.34	14.34	14.34	14.34
Processing Recovery	%	90.0	90.0	90.0	90.0
Plant Throughput	Mtpa
General/Admin	$/t milled	7.40	7.40	7.40	7.40
Gold Price (Resource)	$/oz	1,500	1,500	1,500	1,500
Total Process Cost (per ore tonne mined)	$/t ore	14.34	14.34	14.34	14.34
Total Mining Cost (per ore tonne mined)	$/t ore	33.38	35.82	36.94	35.38
Marginal In-situ Cut-off Grade	g/t Au	0.60	0.60	0.60	0.60
Strip Ratio					10.0
Full Grade Ore In-situ Cut-off Grade	g/t Au	1.45	1.51	1.54	1.50

18^th September 2018 Page 202

Kibali Gold Mine NI 43-101 Technical Report

The Gorumbwa $1,500/oz pit optimisation is not constrained in anyway. The optimisation generates enough significant free cash flow that it would pay for the additional relocation assistance programme (RAP) incurred during mining the pit.

Pakaka

The cut-off grade calculations for the Pakaka open pit Mineral Resource are broken down in Table 14-62.

Table 14-62 Pakaka 2017 Optimisation Parameters

Material Type	Unit	Oxide	Trans	Fresh	Total
Waste Cost	$/t mined	2.72	2.80	2.88	2.80
Extra Ore Cost -GC+Ore-Rehandle + Overhaul	$/t mined	1.38	1.38	1.38	1.38
GC Only	$/t mined	0.75	0.75	0.75	0.75
Dilution	%	10%	10%	10%	10%
Ore Loss	%	3%	3%	3%	3%
Haulage Cost	$/t mined	1.05	1.05	1.05	1.05
Process Cost	$/t milled	14.34	14.34	16.10	14.34
Processing Recovery	%	88.7	81.3	80.2	83.4
Plant Throughput	Mtpa
General/Admin	$/t milled	7.40	7.40	7.40	7.40
Gold Price (Resource)	$/oz	1,500	1,500	1,500	1,500
Total Process Cost (per ore tonne mined)	$/t ore	15.39	15.39	17.15	15.39
Total Mining Cost (per ore tonne mined)	$/t ore	15.33	15.75	16.19	15.76
Marginal In-situ Cut-off Grade	g/t Au	0.64	0.70	0.76	0.68
Strip Ratio					4.1
Full Grade Ore In-situ Cut-off Grade	g/t Au	1.02	1.12	1.20	1.09

Pakaka haulage costs are incorporated into the mining costs as both the haulage and mining are operated by the same contractor.

Geometallurgical work initiated in early 2016 has primarily focused on Pakaka, where feasibility testwork had identified mainly two domains of high and low-grade arsenic domains. With limited metallurgical testwork data available, it was demonstrated that there was:

☐ Direct correlation between gold grade and arsenic content

☐ Inverse correlation between recovery and arsenic grade.

Consequently, the recoveries shown in Table 14-62 are an average for each weathering classification. The detailed Pakaka domain recoveries are detailed in Table 14-18 and the geometallurgical domains are shown in Figure 14-39.

18^th September 2018 Page 203

Kibali Gold Mine NI 43-101 Technical Report

Figure 14-39 Plan View Map of the Pakaka Geometallurgical Domains and Their Spatial Correlation with the Mineralisation Resource Domains

In addition to applying these recoveries to the LOM optimisation, the delineation of the six geometallurgical domains is utilised to optimise the blending strategy during feeding in the plant (Table 14-63). Apart from understanding the recoveries associated with the individual domains, arsenic concentration in the plant feed blend is used to maintain the thresholds (<2,000 ppm) that ensures, not only stable recovery, but reagents consumption.

18^th September 2018 Page 204

Kibali Gold Mine NI 43-101 Technical Report

Table 14-63 Pakaka Geometallurgical Domained Recoveries

Domain	Description	Weathering	BRTs Average     Dissolution     (%)	Arsenic          Assay         (ppm)	Feasibility     Direct Leach  (%)	Comments
1	LG/LAS/LR	SAP	84.1	<1000
		OX TR	86.8	<1000
		RED TR	81.6	<1000
2	HG/HAS/HR	SAP	90.8	>2000
		OX TR	90.4	>2000
		RED TR	86	>2000
3	HG/HAS/LR	FR	75.2	>2000	59.6	Feasibility dissolution exclude gravity, so use BRT value to cater for gravity
4	LG/LAS/HR	SAP	85.5	<1000
		RED TR	92.6	<1000
		FR	93.4	<1000	87.3	Use feasibility number and the BRT number for plant performance tracking
5	MG/MAS/HR	SAP	87.4	1000 - 2000
		FR	88.3	1000 - 2000	87.3	Feasibility split only caters for above and below 0.2% arsenic content. Sample represents below 0.2%
6	HG/HAS/HR	SAP	89	>2000
		OX TR	89.6	>2001
		FR	88.8	>2002

18^th September 2018 Page 205

Kibali Gold Mine NI 43-101 Technical Report

Kombokolo

The cut-off grade calculations for the Kombokolo open pit Mineral Resource are broken down in Table 14-64.

Table 14-64 Kombokolo 2017 Optimisation Parameters

Material Type	Unit	Oxide	Trans	Fresh	Total
Waste Cost	$/t mined	2.65	2.72	2.84	2.74
Extra Ore Cost -GC+Ore-Rehandle + Overhaul	$/t mined	1.19	1.19	1.19	1.19
GC Only	$/t mined	0.75	0.75	0.75	0.75
Dilution	%	10%	10%	10%	10%
Ore Loss	%	3%	3%	3%	3%
Haulage Cost	$/t mined
Process Cost	$/t milled	14.34	14.34	14.34	14.34
Processing Recovery	%	85.0	85.0	85.0	85.0
Plant Throughput	Mtpa
General/Admin	$/t milled	7.40	7.40	7.40	7.40
Gold Price (Resource)	$/oz	1,500	1,500	1,500	1,500
Total Process Cost (per ore tonne mined)	$/t ore	14.34	14.34	14.34	14.34
Total Mining Cost (per ore tonne mined)	$/t ore	28.75	29.53	30.68	29.65
Marginal In-situ Cut-off Grade	g/t Au	0.64	0.64	0.64	0.64
Strip Ratio					9.4
Full Grade Ore In-situ Cut-off Grade	g/t Au	1.20	1.22	1.25	1.22

Pamao

The cut-off grade calculations for the Pamao open pit Mineral Resource are broken down in Table 14-65.

Table 14-65 Pamao 2017 Optimisation Parameters

Material Type	Unit	Oxide	Trans	Fresh	Total
Waste Cost	$/t mined	2.85	2.88	2.95	2.89
Extra Ore Cost -GC+Ore-Rehandle + Overhaul	$/t mined	1.31	1.31	1.31	1.31
GC Only	$/t mined	0.75	0.75	0.75	0.75
Dilution	%	10%	10%	10%	10%
Ore Loss	%	3%	3%	3%	3%
Haulage Cost	$/t mined	1.05	1.05	1.05	1.05
Process Cost	$/t milled	14.34	14.34	16.10	14.34
Processing Recovery	%	90.9	85.0	83.5	86.5
Plant Throughput	Mtpa
General/Admin	$/t milled	7.40	7.40	7.40	7.40
Gold Price (Resource)	$/oz	1,500	1,500	1,500	1,500
Total Process Cost (per ore tonne mined)	$/t ore	15.39	15.39	17.15	15.39
Total Mining Cost (per ore tonne mined)	$/t ore	15.33	15.75	16.19	15.76
Marginal In-situ Cut-off Grade	g/t Au	0.64	0.70	0.76	0.68
Strip Ratio					4.1
Full Grade Ore In-situ Cut-off Grade	g/t Au	1.02	1.12	1.20	1.09

18^th September 2018 Page 206

Kibali Gold Mine NI 43-101 Technical Report

Mengu Village and Marakeke

For Mengu Village and Marakeke, the Mineral Resources were reported at a cut-off grade above a nominal 0.5 g/t Au.

14.15 Mineral Resources Reporting

The Mineral Resource estimates have been prepared according to the of the guidelines Australasian Code for Reporting of Exploration Results, Mineral Resources and Ore Reserves standards and guidelines published and maintained by the Joint Ore Reserves Committee of the Australasian Institute of Mining and Metallurgy and the Australian Institute of Geoscientists and Minerals Council of Australia (the JORC (2012) Code). Randgold has reconciled the Mineral Resources and Ore Reserves to Canadian Institute of Mining, Metallurgy and Petroleum (CIM) 2014 Definition Standards for Mineral Resources and Mineral Reserves dated May 10, 2014 (CIM (2014) Standards) as incorporated with NI 43-101 and there are no material differences.

All reports are generated directly from the modelling software package used. All block model reports are generated using the ‘au_ok’ field for OK models and are combined with other attributes flagged within the model depending on the purpose of the report.

In the case of UC models, different attributes are used to report depending on the purpose of the report. All reports required are completed only within the mineralisation wireframe domains.

The cut-off grade selected for limiting each of the Mineral Resources corresponds to the insitu marginal cut-off grade using a gold price of $1,500/oz.

For the open pit Mineral Resources, the pit shell selected for limiting each of the Mineral Resources corresponds to a gold price of $1,500/oz. As a result of the optimisation process, this pit shell selection will result in the highest undiscounted net present value of the deposit, at $1,500/oz.

Underground Mineral Resources were reported within a minimum mineable stope shape, applying reasonable mineability constraints, including a minimum mining width, a reasonable distance from current or planned development, and a measure of assumed profitability at the related resource cut-off grade, thus deemed as having a reasonable prospect of eventual economic extraction.

The total year end 2017 Kibali Mineral Resource estimate are listed in Table 14-66. The open pit Mineral Resources are estimated within a $1,500 pit shell and at variable economic cut-offs depending on the deposit. The underground Mineral Resources are declared within optimised stopes at a 1.6 g/t Au cut-off and within the bounding box.

The Kibali Measured and Indicated Mineral Resources, as of 31st December 2017, are estimated at 126 Mt at 3.26 g/t Au containing 13 Moz of gold, with an additional Inferred Resource of 44 Mt at 2.8 g/t Au containing 3.3 Moz of gold. This represents a 2% drop in grade and a 6% drop in

18^th September 2018 Page 207

Kibali Gold Mine NI 43-101 Technical Report

tonnes, resulting in an overall 4% decrease in contained gold ounces, relative to the 2016 Mineral Resource estimate.

The Qualified Person is not aware of any environmental, permitting, legal, title, socioeconomic, marketing, metallurgical, fiscal, or other relevant factors, that could materially affect the Mineral Resource estimate.

18^th September 2018 Page 208

Kibali Gold Mine NI 43-101 Technical Report

Table 14-66 Kibali Gold Project Mineral Resource Estimate as of 31st December 2017

Deposit	Cut Off   Grade (g/t   Au)	Measured			Indicated			Measured + Indicated			Inferred
		Tonnes   (Mt)	Grade (g/t   Au)	Contained Au   (Moz)	Tonnes   (Mt)	Grade (g/t   Au)	Contained Au   (Moz)	Tonnes   (Mt)	Grade (g/t   Au)	Contained Au   (Moz)	Tonnes   (Mt)	Grade (g/t Au)	Contained Au   (Moz)
Open Pit
Stockpiles		1.7	1.45	0.08	-	-	-	1.7	1.45	0.08	-	-	-
KCD**	0.63	3.3	2.62	0.28	9.1	2.1	0.62	12	2.24	0.89	6.8	1.8	0.4
Sessenge**	0.66	-	-	-	5.8	2.13	0.4	5.8	2.13	0.4	1.7	1.9	0.1
Sessenge SW	0.5	-	-	-	-	-	-				0.47	1.7	0.026
Pakaka**	0.68	3.4	2.64	0.29	6.1	2.36	0.46	9.5	2.46	0.76	1.1	1.7	0.059
Mengu Hill	0.83	-	-	-	-	-	-				1.6	2.6	0.14
Gorumbwa**	0.6	1.3	2.37	0.097	4.8	3.14	0.49	6.1	2.98	0.59	0.2	3.3	0.024
Megi	0.5	-	-	-	2.3	1.55	0.11	2.3	1.55	0.11	4.6	1.7	0.25
Pamao**	0.66	-	-	-	8.6	1.51	0.42	8.6	1.51	0.42	2.7	1.8	0.15
Kombokolo**	0.64	0.69	3.14	0.069	0.64	3.1	0.063	1.3	3.12	0.13	0.6	2.4	0.045
Mengu Village	0.5	-	-	-	1.1	1.54	0.055	1.1	1.54	0.055	-	-	-
Marakeke	0.5	-	-	-	-	-	-				2.5	1.5	0.12
Rhino	1.16	-	-	-	-	-	-				-	-	-
OP Total		10	2.44	0.81	38	2.11	2.6	49	2.18	3.4	22	1.8	1.3
Underground
KCD UG**	1.6	12	5.57	2.1	65	3.64	7.6	77	3.94	9.8	22	2.8	2
Open Pit + Underground
Total Resources		22	2	3	104	3.07	10	126	3.26	13	44	2.3	3.3

The Mineral Resource estimate has been prepared according to JORC (2012) Code. Kibali have reconciled the Mineral Resources to CIM (2014) Standards, and there are no material differences. All Mineral Resources tabulations are reported inclusive of that material which is then modified to form Ore Reserves.

Open pit Mineral Resources are Mineral Resources within the $1,500/oz pit shell reported at an average cut-off grade of 0.6 g/t Au.

Underground Mineral Resources in the KCD deposit are Mineral Resources, which meet a cut-off grade of 1.6 g/t Au and are reported insitu within a minimum mineable stope shape, at a gold price of $1,500/oz.

Mineral Resources were estimated by Simon Bottoms, CGeol, an officer of the company and Qualified Person.

Numbers may not add due to rounding.

18^th September 2018 Page 209

Kibali Gold Mine NI 43-101 Technical Report

14.16 2017 Versus 2016 Mineral Resource Comparison

Annual comparisons of Mineral Resources are completed to quantify and verify changes due to model change, depletion, and changes due to the cut-off grade, where a calculated 2017 model value is compared to the actual declared 2017 resources. Model changes and depletion at KCD, Sessenge, Gorumbwa, Pakaka, Pamao, and Kombokolo were updated in 2017, with the rest remaining the same as 2016.

KCD Open Pit Resources

Open pit Mineral Resources are reported within 2016 $1,500 pit shell and depleted with December 2017 mined surfaces and reported above the underground reporting box wireframe solid, with varying RL. This varying RL now limits the 5000 lode, 5680 mRL, and 3000 lode reporting to 5682.5 mRL for the 2017 Mineral Resource. This varying RL has been put in place to ensure that the open pit Mineral Resource excludes all material that forms part of the underground Mineral Resource and accordingly, this material is reported as part of the underground Mineral Resource.

Although reporting box change accounts for the majority of the drop in contained gold ounces, model changes have accounted for a significant increase in ounces as well, with a loss in Indicated Mineral Resource, however, a gain in Inferred Mineral Resource. Durba Hill remains a large part of the Inferred Mineral Resource at KCD which represents an opportunity for an increase in tonnes below the current reserve pit.

KCD Underground Resources

The 2017 underground Mineral Resources were reported within optimised minable stope shapes at 1.6 g/t Au and within the underground reporting box wireframe solid, with varying RL. This varying RL now limits the 5000 lode, 5680 mRL, and 3000 lode reporting to 5682.5 mRL for the 2017 Mineral Resource. This varying RL has been put in place to ensure that all material that forms part of the underground Mineral Resource is excluded from the Open pit Mineral Resource. The Mineral Resource is depleted with December 2017 SOX depletion. Results of the reconciliations are presented Table 14-67.

Table 14-67 KCD Underground 2017 vs 2016 Comparison Within Bounding Box and Within MSO

KCD Underground	Tonnes	Grade (g/t Au)	Ounces     (Oz Au)	Comments
2016 Declared resources	100,809,306	3.70	1,992,728
Depletion	(1,787,739)	5.51	(316,791)	SOX depletions as of December, 2016
Model Change	292,072	12.04	113,087
Cut-off Change	-	-	-	Reported from Resources Mineable Shapes
Other Change	-	-	-
2017 Model Calculation	99,313,638	3.69	1,789,025
2017 Declared resources	99,313,638	3.69	1,789,025
2017 Model Calculation vs 2017 Declared Resources	0.00%	0.00%	0.00%

18^th September 2018 Page 210

Kibali Gold Mine NI 43-101 Technical Report

KCD Underground	Tonnes	Grade     (g/t Au)	Ounces       (Oz Au)	Comments
2016 Declared vs 2017	-1%	0%	-2%
Net Change	(1,495,668)	4.24	(203,704)

The 2017 model changes, after depletion, resulted in a 1% decrease in the tonnage and a 2% decrease in the contained gold ounces when the model is compared to the 2016 model. Table 14-68 outlines the results of the changes within the resource categories when depleted to the 2017 resource year.

Table 14-68 KCD Underground 2017 versus 2016 Comparison by Classification

2015 Declared (Depleted to EOY 2015)				2016 MSO (Depleted to EOY 16)			Material Difference
Classification	Tonnes	Grade     (g/t Au)	Contained     Gold (oz)	Tonnes	Grade     (g/t Au)	Contained     Gold (oz)	Tonnes	Ounces
Measured	7,555,886	3.40	825,559	11,933,390	5.57	2,135,978	4,377,504	1,310,419
Indicated	68,257,385	4.10	8,999,750	65,262,331	3.64	7,638,032	(2,995,054)	(1,361,717)
M+I	75,813,272	4.03	9,825,308	77,195,721	3.94	9,774,010	1,382,449	(51,298)
Inferred	24,996,035	2.70	2,167,420	22,117,917	2.83	2,015,014	(2,878,117)	(152,406)

A net change of 1.50 Mt for 203 koz Au has resulted from:

☐ 1.79 Mt at 5.11 g/t Au for 317 koz Au mined out (depletion) in 2017.

☐ Model changes resulting in a gain of 292 kt for 133 koz Au gain.

Sessenge

Table 14-69 presents the Sessenge 2017 versus 2016 Mineral Resource comparison within the $1,500 pit shell.

Table 14-69 Sessenge 2017 vs 2016 Comparison Within $1,500 Pit Shell

Sessenge	Tonnes	Grade     (g/t Au	Ounces     (Oz Au	Comments
2016 Declared resources	6,028,074	1.79	346,915
Depletion	-	-	-
Model Change	2,052,342	2.48	163,650	Increase in Inferred. Tightening of Low-grade domain and new HG domains
Cut-off Change	(584,342)	0.51	(9,587)	Change in cut-off from 0.5 g/t to 0.66 g/t Au
Other Change	-	-	-
2017 Model Calculation	7,496,074	2.08	500,978
2017 Declared resources	7,496,074	2.08	500,978
2017 Model Calculation vs 2017 Declared Resources	0.00%	0.00%	0.00%
2016 Declared vs 2017	24%	16%	44%
Net Change	1,468,000	3.26	154,063

Sessenge remodelling and cut-off grade change resulted in a net gain in tonnage of 1.47 Mt for 154 koz Au for a total declared resource of 7.5 Mt at 2.08 g/t Au for 501 koz Au. The contribution to the net change is summarised below:

☐ Model Change increase of 2.05 Mt for 164 koz Au.

18^th September 2018 Page 211

Kibali Gold Mine NI 43-101 Technical Report

☐ Cut-off change from 0.50 g/t to 0.66 g/t Au results in decrease of 584 kt for 9.6 koz Au.

Gorumbwa

Table 14-70 presents the Gorumbwa 2017 versus 2016 Mineral Resource comparison within the $1,500 pit shell.

Table 14-70 Gorumbwa 2017 vs 2016 Comparison Within $1,500 Pit Shell

Gorumbwa	Tonnes	Grade     (g/t Au)	Ounces     (Oz Au)	Comments
2016 Declared resources	6,156,068	3.09	611,321
Depletion	-	-	-
Model Change	195,310	-0.12	(778)	Top cuts modified
Cut-off Change	(5,530)	1.51	(269)	Increased cut-off from 0.59 g/t to 0.60 g/t Au
Other Change	-	-	-
2017 Model Calculation	6,345,849	2.99	610,274
2017 Declared resources	6,345,849	2.99	610,274
2017 Model Calculation vs 2017 Declared Resources	0.00%	0.00%	0.00%
2016 Declared vs 2017	3%	-3%	0%
Net Change	189,780	-0.17	(1,046)

The model changes represent a minimal decrease in overall contained gold ounces within the $1,500 shell due to:

☐ Additional drilling data resulted in a 3% increase in tonnage and a drop of 1 koz Au.

☐ 3% loss in grade due to change in top cutting reduced the grade from 3.09 g/t to 2.99 g/t Au.

Pakaka

Table 14-71 outlines results for Pakaka 2017 versus 2016 Mineral Resource comparison within the $1,500 pit shell.

Table 14-71 Pakaka 2017 vs 2016 Comparison Within $1,500 Pit Shell

Pakaka	Tonnes	Grade     (g/t Au)	Ounces     (Oz Au)	Comments
2016 Declared resources	15,740,543	2.12	1,070,401
Depletion	(4,596,033)	1.99	(294,491)
Model Change	(359,502)	-3.45	39,912	Material conversion. Increase in measured, but loss in Inferred material
Cut-off Change	(143,476)	0.39	(1,789)	Cut-off change from 0.64 g/t to 0.68 g/t Au
Other Change	-	-	-
2017 Model Calculation	10,641,532	2.38	814,033
2017 Declared resources	10,641,532	2.38	814,033
2017 Model Calculation vs 2017 Declared Resources	0.00%	0.00%	0.00%
2016 Declared vs 2017	-32%	12%	-24%
Net Change	(5,099,010)	1.56	(256,368)

18^th September 2018 Page 212

Kibali Gold Mine NI 43-101 Technical Report

The net resource change at Pakaka is a decrease of 5.1 Mt for 256 koz Au within the $1,500 pit shell, which can be broken down due to:

☐ Mining depletion accounting for decrease of 4.6 Mt at 1.99 g/t Au for 294 koz Au.

☐ Model change resulting in a decrease of 360 kt with an increase of 40 koz Au driven from an increase in grade by 12%.

☐ Cut-off grade change due to change in G&A cost resulted in decrease of 143 kt for 1.8 koz Au.

Kombokolo

Table 14-72 presents the Kombokolo 2017 versus 2016 Mineral Resource comparison within the $1,500 pit shell.

Table 14-72 Kombokolo 2017 vs 2016 Comparison Within $1,500 Pit Shell

Kombokolo	Tonnes	Grade     (g/t Au)	Ounces     (Oz Au)	Comments
2016 Declared resources	2,803,214	2.74	247,181
Depletion	(665,661)	2.85	(61,059)	Pit failure not effect final Year end depletion
Model Change	(207,018)	1.21	(8,082)	Material loss due to updated LG domains and new HG domain
Cut-off Change	(11,035)	0.70	(249)
Other Change	-	-	-	Changed from 0.62 g/t to 0.64 g/t Au
2017 Model Calculation	1,919,499	2.88	177,791
2017 Declared resources	1,919,499	2.88	177,791
2017 Model Calculation vs 2017 Declared Resources	0.00%	0.00%	0.00%
2016 Declared vs 2017	-32%	5%	-28%
Net Change	(883,714)	2.44	(69,390)

The year on year resource, after depletion, shows a decrease of 884 kt for 69 koz Au for Kombokolo. This is driven by the following:

☐ Depletion accounted for 666 kt at 2.85 g/t Au for 61 koz Au decrease.

☐ Model change resulted in a loss of 207 kt for 8 koz Au, due to the creation of a high-grade domain within Kombokolo.

☐ Cut-off change due to G&A produced a loss of 11 kt for 249 oz Au.

18^th September 2018 Page 213

Kibali Gold Mine NI 43-101 Technical Report

Pamao

Table 14-73 outlines the Pamao 2017 versus 2016 Mineral Resource comparison within the $1,500 pit shell.

Table 14-73 Pamao 2017 vs 2016 Comparison Within $1,500 Pit Shell

Pamao	Tonnes	Grade   (g/t Au)	Ounces   (Oz Au)	Comments
2016 Declared resources	14,488,297	1.53	710,447
Depletion	-	-	-
Model Change	(2,593,947)	1.53	(127,960)	Biggest change in Inferred. WF reviewed and LG/Waste material removed
Cut-off Change	(594,073)	0.52	(9,938)	Cut-off change from 0.5 g/t to 0.66 g/t Au
Other Change	-	-	-
2017 Model Calculation	11,300,277	1.58	572,549
2017 Declared resources	11,300,277	1.58	572,549
2017 Model Calculation vs 2017 Declared Resources	0.00%	0.00%	0.00%
2016 Declared vs 2017	-22%	3%	-19%
Net Change	(3,188,020)	1.35	(137,897)

The Pamao Mineral Resource comparison from 2016 versus 2017 shows a 22% loss in tonnes, a 3% increase in grade, and a 19% loss in gold ounces. This is predominantly driven by model change, accounting for 2.6 Mt for 127 koz Au, due to the previously mentioned remodelling to remove the low-grade tail of the distribution.

Mengu Hill

Table 14-74 outlines the Mengu Hill 2017 versus 2016 Mineral Resource comparison within the $1,500 pit shell.

Table 14-74 Mengu 2017 vs 2016 Comparison Within $1,500 Pit Shell

Mengu Hill	Tonnes	Grade   (g/t Au)	Ounces   (Oz Au)	Comments
2016 Declared resources	2,137,148	2.73	187,919
Depletion	(487,451)	3.11	(48,817)
Model Change	-	-	-
Cut-off Change	(10,925)	0.67	(235)	Change in cut-off from 0.82 g/t to 0.83 g/t Au
Other Change	-	-	-
2017 Model Calculation	1,638,773	2.64	138,868
2017 Declared resources	1,638,773	2.64	138,868
2017 Model Calculation vs 2017 Declared Resources	0.00%	0.00%	0.00%
2016 Declared vs 2017	-23%	-4%	-26%
Net Change	(498,376)	3.06	(49,051)

18^th September 2018 Page 214

Kibali Gold Mine NI 43-101 Technical Report

14.17 Reconciliation

Kibali has a standard weekly, end of month (EOM) and end of quarter production measurement system that reports and provides reconciliation between grade control and the monthly mine production.

The Kibali measurement system tracks daily, weekly, monthly, quarterly and year to date production grade control results versus the Plant. The system tracks both underground and open pit domains production against the block model. Summary reports are prepared weekly, monthly, and quarterly.

The reconciliation between GC call and Plant check out showed a good reconciliation during 2017. Some issues have been investigated and fixed within the year, but overall these have had a negligible impact on the mine to mill reconciliation (Table 14-75).

Table 14-75 Kibali Mine Call Factor (MCF) 2017 EOY Reconciliation

Dept	Recon Ore Mine, Stockpiles and Plant Out	Year End 2017
		Tons	Grade	Ounces
GC	Mine	6,773,574	3.21	699,740
GC	Stockpile Change	-1,171,800	1.42	-53,312
GC	GC Actual Feed	7,909,415	2.39	744,588
GC	Scats Stock Change	66,352	3.27	6,969
Plant	Cone Change	13,713	3.27	1,444
GC	GC Call	7,829,349	2.92	736,175
GC	Plant Check Out	7,618,932	2.92	715,950
GC vs Plant	MCF (%) GC Call vs Plant Check Out	97	100	97

Figure 14-40 presents a chart of the Kibali Mine production with weekly feed source ratio versus pulp call versus gold after smelting. As shown Figure 14-41 there was a gain on grade observed in December which relates to an undercall of grade from two KCD underground stopes (XC22 and XC16), and in-plant tank & sump sludge was fed through the circuit from week 49 to week 51.

18^th September 2018 Page 215

Kibali Gold Mine NI 43-101 Technical Report

Figure 14-40 2017 Kibali Mine Production with Weekly Feed Source Ratio versus Pulp Call versus Gold after Smelting

Figure 14-41 2017 Weekly Grades Comparison (GC Call Grade vs Plant Check Out Grade vs Carbon Loading)

As shown in Figure 14-42, during weeks 23 and 36 a tonnage bias was present between GC call and Plant Check out tonnes. This was attributed to changes in the broken and crushed material densities applied as a function of changes in drill and blast practice resulting in different fragmentation rates. This was resolved by performing a quarterly broken density measurement on all sources using a weigh bridge for mass determination and laser scanner for volumetric determinations.

18^th September 2018 Page 216

Kibali Gold Mine NI 43-101 Technical Report

Figure 14-42 2017 Weekly Tonnage Comparison (GC Call Tonnes vs Plant Check Out Tonnes)

14.18 Discussion

External Resource Audits

An independent audit was undertaken in 2012 on the Mineral Resource estimate by Quantitative Group (QG). The audit focussed primarily on KCD due to its dominant size. The results of the audit set out some minor recommendations to improve QA/QC compliance, sampling procedures, and modelling methodologies, which have been acted upon.

However, the audit failed to identify a bias in drill direction of the sub vertical underground lodes and, as such, the risk to the resource model was not fully appreciated until 2016 when first advance grade control results, drilled from underground development, were significantly different to that which had been modelled in the block model.

Subsequently, an additional full resource audit by Optiro was completed in 2017, after the majority of the underground lodes were covered with an initial pass of underground resource definition drilling. The QP considers that the updated 2017 Mineral Resource estimation fully accounts for the differences in grade previously identified 2016 and does not show a material difference between drill results and estimated grades.

Optiro conducted an independent audit of the Mineral Resource estimates for the Project in the second half of 2017 (Optiro, 2017). The audit focused on two aspects:

1. Kibali Mineral Resource site procedure and process review.

2. Mineral Resource estimation validation.

Optiro concluded that the Mineral Resource estimation processes used by Kibali Goldmines are considered, by Optiro, to be at a level commensurate with industry best practice based on the process review completed in August 2017 (Optiro, 2017).

18^th September 2018 Page 217

Kibali Gold Mine NI 43-101 Technical Report

Following a second phase of Mineral Resource estimation validation, Optiro concluded that:

“Optiro endorses the Mineral Resources and Ore Reserves to be declared for Kibali as at 31st December 2017. Optiro endorses the basis for the mineable resource declaration as being compliant with the principles of both the JORC and SAMREC Codes. The processes underlying the generation and declaration of Mineral Resources and Ore Reserves reflect good to best world practice.”

The application of optimised mineable resource shapes applies reasonable mineability constraints, including a minimum mining width, a reasonable distance from current or planned development, and a measure of assumed profitability at the related resource cut-off grade. This change in reporting method has removed isolated areas of mineralisation and lowered the grade of the reported underground resource by reporting all material, geologically classified as ore within each mineable shape, whilst ensuring the overall shape meets the resource cut-off grade. Thereby ensuring that the Mineral Resources are reported in line with industry best practise with specific regard to underground resources only being reported if there is an intention to mine the material.

Optiro acknowledged that the “estimation of resources at Kibali is complex, with a number of very large models. Randgold has tackled the estimation in a systematic way, with largely common approaches to compositing, top cuts, declustering, kriging neighbourhood analysis (KNA), estimation parameters, classification, and validation. The documentation of the estimation and validation is generally very comprehensive. The processes follow good to best industry practice.”

As part of the process review, Optiro has made a number of suggestions to Kibali Goldmines for continual improvement and to enhance and streamline the current systems.’

The following findings are summarised from the resource process review. No essential issues (issues to be addressed before endorsement can be made) were identified by Optiro. The following recommendations were identified:

☐ The RC subsampling practice needs to be revised to ensure consistent sample weights within the tolerance (3 kg to 3.5 kg) required by the SGS Doko assay laboratory. This will entail changes to procedures and the purchase of a new set of Gilson splitters, which will ensure an unbiased and consistent subsample. This process has already commenced during the site visit.

☐ In line with the above, once RC samples of the right mass are being delivered to the laboratory they should all be processed through the Boyd crusher and associated rotary sample divider. The laboratory should not need to split the samples using a riffle splitter as is current practice. This process has already commenced during 2017 shortly after the site visit.

☐ Blank samples (granite fragments) should be selectively inserted within the high-grade intersections as logged, whether for diamond core or RC chips. This will ensure maximum effectiveness and value for money of the blank insertion process. This process has already commenced during 2017 shortly after the site visit.

☐ Increase the use of implicit modelling software for lithology and structural modelling.

18^th September 2018 Page 218

Kibali Gold Mine NI 43-101 Technical Report

☐ Generation of more formal guidelines for the relative interaction of top cutting and high yield restriction

Since the audit, the items regarding sample quality were addressed and the recommendations were implemented.

Previously, the use of implicit modelling has predominantly been completed offsite. Going forward into 2018, the onsite geologists will be trained to increase the use of the intrinsic modelling, prominently for lithology and structural analysis.

For resource estimation, improved guidelines for the relationship between top cutting and high yield restriction are to be reviewed and guidance to be issued. The comparison of block model grades to declustered capped composite data has been implemented since the auditor’s site visit.

Any issues that were highlighted during the review process were communicated to Kibali Goldmines from Optiro. The issues were reviewed and revised models issued, where applicable. From the issues highlighted, none of them are deemed to have a material impact on the Kibali Mineral Resource Estimations.

As part of the review, Optiro has made a number of ‘Value Add’ suggestions to Kibali Goldmines for continual improvement and to the enhancement of the Kibali Mineral Resource standards:

Optiro recommendations are as follows:

☐ Vulcan dynamic anisotropy modelling to be investigated, if models are acceptable it would reduce the need for estimation domains due to orientation change.

☐ Summary of basis of the multiple search passes to be clearly documented – i.e. 95% of the sill.

☐ A tabulation of the summary statistics of gold grades of the residual composite lengths and the composite lengths greater than length threshold chosen.

All of the value add suggestions are due to be incorporated into the next round of 2018 Mineral Resource updates.

Relative Accuracy / Confidence of the 2017 Mineral Resource Estimate

The QP’s offer the following conclusions regarding the relative accuracy / confidence of the 2017 Mineral Resource Estimate:

☐ The application of optimised resource shapes applies reasonable mineability constraints including a minimum mining width, a reasonable distance from current or planned development, and a measure of assumed profitability at the related resource cut-off grade. This change in reporting method has removed isolated areas of mineralisation and lowered the grade of the reported underground resource by reporting all material, geologically classified as ore, within each mineable shape, whilst ensuring the overall shape meets the resource cut-off grade. Thereby ensuring that the Mineral Resources are

18^th September 2018 Page 219

Kibali Gold Mine NI 43-101 Technical Report

reported in line with industry best practise with specific regard to underground resources only being reported if there is an intention to mine the material.

☐ In 2017, Optiro completed a resource audit at Kibali which included the KCD underground model in line with comments and recommendations in the 2016 resource report. This is due to the substantial advanced grade control drilling completed to understand the bias in drill direction of the sub vertical underground lodes.

☐ Optiro acknowledges that the estimation of resources at Kibali is complex, with a number of very large models. Kibali Goldmines has tackled the estimation in a systematic way, with largely common approaches to compositing, top cuts, declustering, KNA, estimation parameters, classification, and validation. The documentation of the estimation and validation is generally very comprehensive. The processes follow good to best industry practice.

☐ The KCD underground Mineral Resource and geological model has been significantly affected by a sampling bias with the feasibility drilling data being conducted from surface. This has meant that some of the sub vertical lodes such as 9105 and 5101 were initially delineated using sub optimal drill directions. Since 2016, there has been a significant quantity of Advanced Grade control drilling conducted from the underground development – where the drilling could be completed with perpendicular angles of intersection to the primary 9105, 5101, and 5110 ore lodes. The results of this drilling have significantly improved the modelled definition of the banded ironstone as the marker unit for the km-scale NE plunging fold structure, which acts as the primary control on the positioning of the 5000 and 9000 ore lodes and delineated zones of internal waste within the 9105. This has resulted in a significant model change affecting both the 2017 Mineral Resources and Ore Reserves.

☐ Significant additional advanced grade control drilling will continue to be conducted throughout 2018. It is anticipated that this drilling will define further model changes in areas not yet tested, particularly in the lower portion and down plunge of the 9105 lode, although additional work since the 2016 resource model has confirmed the updated model specifically in the upper portions of the 5101 and 5110 lodes. However, further model changes have been seen in the down plunge portions of the 5101 which have not previously included any advanced grade control drilling. As at the Mineral Resource cut-off date, all material model changes are incorporated; however further changes were reported as of EOM November grade control model. These changes will be incorporated into the 2018 Mineral Resource and have already been factored in within the 2017 Ore Reserve.

☐ The 2016 Mineral Resource report highlighted the risk to the 9101 resource due to the lack of additional drilling since completion of the feasibility, but the angle of intersection was favourable, being near perpendicular. During 2017, advanced grade control drilling was completed and confirmed the mineralisation of 9101 and interpreted it to join the 9105, based on the folded BIF model. Further infill grade control drilling is planned for 2018 to more accurately define the local grade variability prior to commencement of mining the 9101 in 2019.

The QP is not aware of any environmental, permitting, legal, title, taxation, socio-economic, marketing, political, or other relevant factors that could materially affect the Mineral Resource estimate.

In the QP’s opinion, the 2017 Kibali Mineral Resource estimate is appropriate for the conversion to Ore Reserves.

18^th September 2018 Page 220

Kibali Gold Mine NI 43-101 Technical Report

15 Ore Reserve Estimate

15.1 Summary

The Ore Reserve estimates have been prepared according to the guidelines Australasian Code for Reporting of Exploration Results, Mineral Resources and Ore Reserves standards and guidelines published and maintained by the Joint Ore Reserves Committee of the Australasian Institute of Mining and Metallurgy and the Australian Institute of Geoscientists and Minerals Council of Australia (the JORC (2012) Code). Randgold has reconciled the Mineral Resources and Ore Reserves to Canadian Institute of Mining, Metallurgy and Petroleum (CIM) 2014 Definition Standards for Mineral Resources and Mineral Reserves dated May 10, 2014 (CIM (2014) Standards) as incorporated with NI 43-101 and there are no material differences.

As of 31^st December 2017 (100% basis), the total Proved and Probable Ore Reserves in open pits, underground, and stockpiles is estimated to be 66 Mt at an average grade of 4.1 g/t Au, containing approximately 8.7 Moz of gold.

The Ore Reserve has been estimated from the Measured and Indicated Mineral Resources and does not include any Inferred Mineral Resources. The estimate uses updated economic factors, the latest Mineral Resource and geological models, geotechnical and hydrological inputs, and metallurgical processing and recovery updates. The Qualified Person responsible for estimating the Ore Reserves has performed an independent verification of the block model tonnes and grade, and in his opinion the process has been carried out to industry standards.

For the open pit Ore Reserves, economic pit shells were generated using the Lerch-Grossman algorithm within Whittle software and then used in the open pit mine design process and Ore Reserve estimation.

For the KCD underground mine, the Datamine Mineable Shape Optimiser (MSO) was used to evaluate the geological block model to create overall mining shapes. Preliminary stope wireframes were created and planned dilution was added to the mineable stope shape. Datamine’s EPS Scheduler software was used to estimate the diluted mined tonnes, grade, and contained metal of the Ore Reserves. Stopes with a diluted grade below the cut-off grade (2.49 g/t Au) were excluded from the Ore Reserves.

A financial model was constructed to demonstrate that the Ore Reserves are economically viable.

The Total Ore Reserve estimate at 31st December 2017 for the Kibali Mine is summarised in Table 15-1.

18^th September 2018 Page 221

Kibali Gold Mine NI 43-101 Technical Report

Table 15-1 Total Ore Reserve Estimate at December 31, 2017

Type	Category	Tonnes	Attributable   Tonnes*	Grade	Contained   Gold	Attributable   Gold*
		(Mt)	(Mt)	(Au g/t)	(Moz)	(Moz)
Stockpiles	Proved	1.73	0.78	1.45	0.08	0.04
Open Pits	Proved	4.89	2.20	2.72	0.43	0.19
	Probable	16.28	7.33	2.28	1.19	0.54
Underground	Proved	12.37	5.56	4.97	1.98	0.89
	Probable	30.79	13.85	5.06	5.01	2.25
Total Ore Reserves	Proved and Probable	66.05	29.72	4.09	8.68	3.91

*Attributable Gold (Moz) refers to the quantity attributable to Randgold based on Randgold’s 45% interest in the Kibali Gold Mines. Ore Reserves are reported on a 100% and attributable basis.

The Ore Reserve estimate has been prepared according to JORC (2012) Code. The Qualified Person has reconciled the Ore Reserves to CIM (2014) Standards, and there are no material differences.

Open pit Ore Reserves are reported at a gold price of $1,000/oz, except for the KCD pit at $1,100/oz, and an average cut-off grade of 1.0 g/t Au including dilution and ore loss factors.

Underground Ore Reserves are reported at a gold price of $1,000/oz and a cut-off grade of 2.5 g/t Au including dilution and ore loss factors.

Open pit and underground Ore Reserves were estimated by Mr. Andrew Law, HND(MMin), MBA, FAusIMM (CP), FIQA, MAICD, AFAIM, an external consultant and Qualified Person.

Numbers may not add due to rounding.

The year end 2017 Ore Reserve estimate shows a net reduction of 0.49 Moz when compared to the estimate for year end 2016. This is due mainly to mining depletion compensated by some positive model changes resulting from infill grade control drilling and various adjustments to the economic parameters.

The Qualified Person has performed an independent verification of the block model tonnes and grade, and in their opinion, the process has been carried out to industry standards.

The Qualified Person is not aware of any environmental, legal, title, socioeconomic, marketing, mining, metallurgical, infrastructure, permitting, fiscal, or other relevant factors that could materially affect the Ore Reserve estimate.

15.2 Ore Reserve Estimation Process

Resource Models

The Ore Reserve estimate for the open pits uses the block models prepared by the QP responsible for Mineral Resource estimation. The KCD and Sessenge deposits, which had a combined block mode previously, were separated into individual models in 2017. As KCD, Pakaka, and Kombokolo are active mining pits, the block models were depleted with the end of year pit surveys.

Four main mineralised zones, 5101, 5102, 9101, and 9105, comprise the bulk of the underground Ore Reserve (Figure 15-1). Five other mineralised zones - 3101, 3102, 5104, 5105, and 5110 contribute the remaining 12% of the Ore Reserve. The mineralised zones have been modelled as part of the KCD block model, used for both the underground and open pit Ore Reserve estimation.

18^th September 2018 Page 222

Kibali Gold Mine NI 43-101 Technical Report

Figure 15-1 KCD Underground Mining Zones

Open Pits

The estimation of Kibali open pit Ore Reserves is based on the following key inputs:

☐ Mineral Resource models (Ordinary Kriging methodology for all the Kibali Ore Reserve deposits) for the estimated gold content and material weathering type.

☐ Estimated processing and G&A costs.

☐ Metallurgical recovery by material type and by deposit.

☐ Geotechnical wall angle parameters.

☐ KMS (mining contractor) 2017 pricing which was used for mining costs.

☐ Cut-off grade analysis using final estimated costs derived from pit designs and pit schedules, and finalised processing and administration costs. These are based on a $1,000/oz gold price for all the pits except the KCD pit which was run on a $1,100/oz gold price above an open pit underground interface level at the 5685 mRL.

The Open Pit Ore Reserves were estimated as follows:

☐ Open pit stockpiles estimated as of 31st December 2017.

☐ Measured and Indicated Mineral Resources only were used for conversion, with no Inferred Mineral Resources considered.

☐ Depletion of the KCD, Kombokolo, and Pakaka pits block models with end of year actual survey face position as of 31st December 2017.

☐ Use of an integrated mine and feed schedule.

The Kibali open pit Ore Reserves estimate (100% basis) as of 31st December 2017 is presented in Table 15-2.

18^th September 2018 Page 223

Kibali Gold Mine NI 43-101 Technical Report

Table 15-2 Kibali Open Pit Ore Reserves as of 31st December 2017

Category	Tonnes   (Mt)		Grade   (g/t)		Ounces (Moz)
Open Pit Stockpile
Proved		1.73		1.45	0.08
Total		1.73		1.45	0.08
KCD Pit
Proved		1.23		2.27	0.09
Probable		3.71		2.02	0.24
Total		4.94		2.08	0.33
Kombokolo Pit
Proved		0.53		3.45	0.06
Probable		0.49		3.32	0.05
Total		1.02		3.39	0.11
Pakaka Pit
Proved		2.27		2.91	0.21
Probable		1.52		2.12	0.10
Total		3.79		2.59	0.32
Pamao Pit
Proved		-		-	-
Probable		3.52		1.70	0.19
Total		3.52		1.70	0.19
Sessenge Pit
Proved		-		-	-
Probable		3.78		2.45	0.30
Total		3.78		2.45	0.30
Gorumbwa Pit
Proved		0.87		2.43	0.07
Probable		3.25		2.91	0.37
Total		4.12		2.81	0.37
Open Pits and Stockpile
Proved		6.63		2.39	0.51
Probable		16.27		2.28	1.19
Total		22.90		2.31	1.70

Ore Reserves are reported on a 100% basis.

The Ore Reserve estimate has been prepared according to JORC (2012) Code. The Qualified Person has reconciled the Ore Reserves to CIM (2014) Standards, and there are no material differences.

Open pit Ore Reserves are reported at a gold price of $1,000/oz, except for the KCD pit at $1,100/oz, and an average cut-off grade of 1.0 g/t Au including dilution and ore loss factors.

Open pit Ore Reserves were estimated by Mr. Andrew Law, HND(MMin), MBA, FAusIMM (CP), FIQA, MAICD, AFAIM, an external consultant and Qualified Person.

Numbers may not add due to rounding.

The following factors are updated in the 31st December 2017 open pit Ore Reserve update compared to the previous estimate:

☐ Depletion of Ore Reserves with the open pit mined shape.

☐ Infill grade control drilling resulting in resource model changes.

☐ Economic changes to the G&A costs.

☐ A higher gold price used for the KCD.

☐ Depletion of stockpiles.

The net change between the 2016 Ore Reserve estimate and 2017 Ore Reserve estimate has been a decrease of approximately 200koz Au (-10%)

The Ore Reserve changes are summarised in Table 15-3

18^th September 2018 Page 224

Kibali Gold Mine NI 43-101 Technical Report

Table 15-3 Open Pit Ore Reserve Comparison to Previous Estimate

Change	Tonnes (Mt)	Au Grade (g/t)	Au Ounces (Moz)
December 2016 Ore Reserve Estimate	29.3	2.07	1.95
Depletion of 2016 Ore Reserve	-4.75	2.43	-0.37
Model Changes - Infill GC drilling	-0.06	-99.86	0.18
Economic/Cut-off Grade Changes	-0.72	1.31	-0.03
Gold Price Changes	1.03	1.11	0.04
Stockpile Changes	-1.17	1.42	-0.05
December 2017 Ore Reserve Estimate	22.6	2.32	1.7

Underground Mine

The 2017 Ore Reserves estimation process for the KCD underground mine was undertaken by the Kibali site mining technical team and Andrew Fox, Kenmore Mine Consulting Pty Ltd, and has been verified and endorsed by the Qualified Person.

The block models used were sub-cell block models. The geological zones (including mineralised zones) were defined by three dimensional wireframe solids and surfaces. Both the block models and wireframes were created in Maptek Vulcan by the Kibali Goldmines geological team. For further information on the resource block models, refer to appropriate sections within this report. The block models and wireframes were converted to a Datamine format prior to the Ore Reserve estimation using Datamine 5D planner software.

The process undertaken for estimation of the 2017 Ore Reserves was as follows:

☐ 2017 actual and LOM planned costs were reviewed to determine cut-off grades.

☐ Datamine Mineable Shape Optimiser (MSO) was used to evaluate the geological block model mineralisation and determine the areas to be included and the overall mining shapes as a guideline. The MSO shapes have not been used for the Ore Reserve estimate.

☐ Stope section strings (5 m or 10 m interval between sections) were manually created to follow geological block model mineralisation above cut-off grade, using the MSO shapes as a guide. Strings were based on level intervals determined in the previous Ore Reserve estimate.

☐ Planned dilution was included in the stope shape to create a mineable stope shape.

☐ Preliminary stope wireframes were created from the strings.

☐ Preliminary stope wireframes were cut by the parts of the as-built mined solids that intersected them to remove development drives and already-mined parts of stopes.

☐ The total stope tonnes and gold metal were calculated by evaluating the stope wireframes against the block model.

☐ Diluted mined tonnes, grades, and contained metal were calculated in Datamine EPS Scheduler. This included unplanned dilution added as a varying percentage depending on the hanging wall exposure, stope sequence (primary, secondary, advancing transverse, or longitudinal), and the number of paste fill exposures. Ore loss was subtracted as a percentage from the diluted tonnes and contained metal.

☐ Panels with a diluted grade below cut-off were excluded from the Ore Reserve estimate.

18^th September 2018 Page 225

Kibali Gold Mine NI 43-101 Technical Report

☐ Classification of the Ore Reserve was allocated as per the Mineral Resource, and, where necessary, adjusted.

The Kibali underground Ore Reserves estimate (100% basis) as at 31st December 2017 is detailed and presented Table 15-4 and Table 15-5 according to two breakdowns: by classification, zone, and mining method. The conversion of Measured + Indicated Mineral Resources to Ore Reserves is 71%. Only Measured and Indicated Mineral Resources have been included in the Ore Reserve. No Inferred Mineral Resources were considered in the estimate. Only Measured Mineral Resource has been reported as Proved Ore Reserve. The location of Proven and Probable Ore Reserves is illustrated in Figure 15-2.

Figure 15-2 Kibali Underground Ore Reserve Classification (View NW)

18^th September 2018 Page 226

Kibali Gold Mine NI 43-101 Technical Report

Table 15-4 Kibali Underground Ore Reserves by Classification and Zone – 31st December 2017

Category/Zone	Mt	Au g/t	Au Moz
Proved
5101	4.07	5.72	0.75
5102	5.21	4.77	0.80
5105	0.47	4.30	0.07
5110	0.08	2.62	0.01
9105	2.54	4.38	0.36
Total Proved	12.37	4.97	1.98
Probable
3101	3.69	4.36	0.52
3102	0.56	3.61	0.06
5101	8.51	5.63	1.54
5102	2.37	5.08	0.39
5104	0.78	4.66	0.12
5105	0.31	3.82	0.04
9101	9.79	5.25	1.65
9105	4.78	4.48	0.69
Total Probable	30.79	5.06	5.01
Total UG Proved and Probable Reserves	43.15	5.03	6.98

Ore Reserves are reported on a 100% basis.

The Ore Reserve estimate has been prepared according to JORC (2012) Code. The Qualified Person has reconciled the Ore Reserves to CIM (2014) Standards, and there are no material differences.

Underground Ore Reserves are reported at a gold price of $1,000/oz and a cut-off grade of 2.5 g/t Au including dilution and ore loss factors.

Underground Ore Reserves were estimated by Mr. Andrew Law, HND(MMin), MBA, FAusIMM (CP), FIQA, MAICD, AFAIM, an external consultant and Qualified Person.

Numbers may not add due to rounding.

Table 15-5 Kibali Underground Ore Reserves by Classification and Zone – 31st December 2017

Stope method	UG Ore Reserves
	Mt	Au g/t	Au Moz
Transverse Primary	12.01	5.03	1.94
Transverse Secondary	17.29	5.15	2.86
Longitudinal	4.06	4.05	0.53
Transverse Advancing face	9.79	5.25	1.65
Total UG Ore Reserves	43.15	5.03	6.98

Ore Reserves are reported on a 100% basis.

The Ore Reserve estimate has been prepared according to JORC (2012) Code. The Qualified Person has reconciled the Ore Reserves to CIM (2014) Standards, and there are no material differences.

Underground Ore Reserves are reported at a gold price of $1,000/oz and a cut-off grade of 2.5 g/t Au including dilution and ore loss factors.

Underground Ore Reserves were estimated by Mr. Andrew Law, HND(MMin), MBA, FAusIMM (CP), FIQA, MAICD, AFAIM, an external consultant and Qualified Person.

Numbers may not add due to rounding.

The reporting of Proved Ore Reserves underground for the first time is a consequence of infill grade control drilling, ore development providing exposures for mapping in the areas of Measured Mineral Resource, and processing of 4.3 Mt of underground ore to date, providing higher confidence in the Mineral Resource and the mining and processing modifying factors.

During 2017 a significant amount of resource and grade control drilling has been undertaken in the Kibali underground mine. This has led to reinterpretation of the geological setting and a slight reduction in Ore Reserves in existing areas (excluding new areas converted from Inferred Mineral Resources). New areas have been converted from Inferred Mineral Resource to Indicated Mineral Resources (and hence Probable Ore Reserves) in the upper 9101 and lower 9105 zones.

18^th September 2018 Page 227

Kibali Gold Mine NI 43-101 Technical Report

These changes have been considered in the 31st December 2017 Ore Reserve. The net effect is an increase in the Ore Reserve after depletion of 170 koz Au (+2%). The Ore Reserve changes are summarised in Table 15-6. The history of reported Ore Reserves and Measured + Indicated Mineral Resources for the underground mine is shown in Figure 15-3.

Table 15-6 Underground Ore Reserve Comparison to Previous Estimate

Change	Tonnes (Mt)	Au Grade (g/t)	Au Ounces (Moz)
December 2016 Ore Reserve Estimate	41.41	5.35	7.13
Depletion of 2016 Ore Reserve	-1.79	5.50	-0.32
5102 Secondary Stopes - Increase Ore Loss Provision	-0.18	5.20	-0.03
9101 Up Plunge - Ore Reserve Addition	2.26	4.90	0.35
9105 below 5240 mRL - Ore Reserve Addition	0.61	4.60	0.09
3101 Up Plunge - Ore Reserve Addition	0.45	4.60	0.07
Resource Model Changes in Existing Ore Reserve Areas	0.39	-23.99	-0.30
December 2017 Ore Reserve Estimate	43.15	5.03	6.98

Figure 15-3 History of Kibali Underground Measured + Indicated Mineral Resources and Ore Reserves

Ore Stockpiles

Details of surface stockpiles of ore sourced from the open pits are presented in Table 15-7 for reference purposes. There was no underground ore stockpiled on the surface at 31st December 2017, as all underground ore is direct feed to the crusher.

18^th September 2018 Page 228

Kibali Gold Mine NI 43-101 Technical Report

Table 15-7 Kibali Surface Stockpile Ore Reserve as of 31 December 2017

Location	Actual
	Tonnes (kt)	Au (g/t)	Au (koz)
ROMPAD_FGO	104	2.35	7.9
REHANDLE_MO	1,622	1.39	72
Total Ore SP excl. Scats	1,726	1.45	80

Stockpiles are reported on a 100% basis.

The Ore Reserve estimate has been prepared according to JORC (2012) Code. Randgold has reconciled the Ore Reserves to CIM (2014) Standards, and there are no material differences.

Stockpiles were estimated by Mr. Andrew Law, HND(MMin), MBA, FAusIMM (CP), FIQA, MAICD, AFAIM, an external consultant and Qualified Person.

Numbers may not add due to rounding.

15.3 Geotechnical and Hydrogeological Considerations

Open Pit Operations

Void Modelling

Geotechnical engineering for KCD in 2017 was focused on the void modelling of Pushback 3. A site-specific void mapping and scanning exercise was carried out for this modelling.

A specialist underground cavity survey surveying company, 3D Mine Surveying International Limited (3DMSI) was contracted to carry out the survey and prepare the 3D model by using scanners to derive the 3D model of the voids and allowing a spatial image of the pillars. Historical information from SOKIMO was also used before final void modelling was complete (Figure 15-4). The voids model sits within the pushback of the KCD pit.

Figure 15-4 KCD Pushback 3 Voids Model Within the Planned Pit

18^th September 2018 Page 229

Kibali Gold Mine NI 43-101 Technical Report

An SOP has been developed to manage mining around voids where personnel and equipment are exposed to areas of higher risk associated with instability from sub-surface excavations.

Phase 1 of the Gorumbwa historical void scanning was completed by 3DMSI to map the pillar dimensions, integrity, and location of the voids. Results confirmed the extension of voids already picked up from advanced grade control and infill drilling (Figure 15-5).

Figure 15-5 Gorumbwa Sections of Completed Voids

Slope Angles

Results from geotechnical drilling also indicated shallower weathering profiles for the KCD Pushback 3 pit. Three dominant domains were identified and generated based on the rock properties (Figure 15-6).

Appropriate slope angles were then designed for the various domains. The inter-ramp angles provided were then flattened by a few degrees in the Whittle pit shell generations to accommodate the ramps in the final pit designs (Table 15-8).

18^th September 2018 Page 230

Kibali Gold Mine NI 43-101 Technical Report

Figure 15-6 KCD Pushback 3 Geotechnical Domains

Table 15-8 KCD Geotechnical Geometry

Domain	From	To	Bench   Height   (m)	Berm   Width   (m)	Batter   Angle   (°)	Inter Ramp   Slope Angle (°)	Design Consideration
CB1	Surface	5880	5	4	40	27	Weathered and Weathered Shale
	5880	5810	10	5	65	48	Transition to Fresh
CB2	Surface	5880	5	4	40	27	Weathered Shale
	5880	5810	5	4	40	27	Weathered Shale
CB3	Surface	5860	5	4	40	27	Weathered Shale
	5860	5810	5	5	50	30	Weathered Material

Dempers & Seymour Pty Ltd (D&S) was commissioned by Kibali Goldmines to undertake the pit slope design for the Sessenge Pit. A 3D Mining Rock Mass Model (MRMM) was constructed based on geotechnical logging of drill core undertaken at Kibali (Table 15-9). Based on the above mass model, five distinctive geotechnical domains were identified and defined for the Sessenge Pit (Figure 15-7).

Rigorous analyses, including Rock bridge/structure failure criteria for each rock type per geotechnical domain, were completed and pit slope designs excluding haul ramps were recommended (Table 15-10). Slopes were considered as dry slopes and that the necessary dewatering would take place timely as scheduled.

18^th September 2018 Page 231

Kibali Gold Mine NI 43-101 Technical Report

Table 15-9 Rock Mass Properties for the Sessenge Pit

Rock Unit	Rock Strength	Joint Condition	Fracture     Frequency	RMR	MRMR
Weathered	1 MPa -5 MPa	Smooth and Undulating with Soft   Sheared Fine Infill	>40 frac/m	8 - 12	7 - 10
			Spacing <0.0.2 m	Average 11	Average 9
Transition	25 MPa - 50 MPa	Smooth and Undulating with No   Infill (Clean)	7 frac/m	9 - 56	7 - 45
			Spacing 0.15 m	Average 39	Average 31
MCP	100 MPa -   130 MPa	Smooth and Undulating with No   Infill (Clean)	1 frac/m	62 - 77	50 - 62
			Spacing 1.0 m	Average 69	Average 56
CHS	100 MPa -   130 MPa	Smooth and Undulating with No   Infill (Clean)	1 frac/m	60 - 71	48 - 57
			Spacing 1.0 m	Average 65	Average 52
ORE 9001	100 MPa -   130 MPa	Rough and Undulating with No   Infill (Clean)	1 frac/m	61 - 78	49 - 63
			Spacing 1.0 m	Average 66	Average 53
ORE 9003	100 MPa -   130 MPa	Rough and Undulating with No   Infill (Clean)	1.5 frac/m	60 - 75	48 - 61
			Spacing 0.67 m	Average 66	Average 53
SCH	100 MPa -   130 MPa	Rough and Undulating with No   Infill (Clean)	1.2 frac/m	60 - 70	48 - 57
			Spacing 0.8 m	Average 64	Average 52
BIF	100 MPa -   130 MPa	Rough and Undulating with No   Infill (Clean)	0.4 frac/m	68 - 78	55 - 63
			Spacing 2.5 m	Average 73	Average 59

Figure 15-7 Sessenge Geotechnical Domains

18^th September 2018 Page 232

Kibali Gold Mine NI 43-101 Technical Report

Table 15-10 Sessenge Slope Design

Domain	Material	From RL	To RL	Batter Height   (m)	Berm Width   (m)	Batter Angle   (°)	Inter Ramp   Slope (°)
West	Weathered	Surface	5840	5	4	50	50
West	Transition	5840	5830	10	5	55	55
West	Fresh	5830	5790	10	5	75	57
NW	Weathered	Surface	5830	5	4	50	36
NW	Transition	5830	5820	10	5	55	55
NW	Fresh	5820	5790	10	5	75	59
NE	Weathered	Surface	5850	5	4	50	35
NE	Transition	5850	5830	10	5	55	47
NE	Fresh	5830	5790	10	5	75	57
SW	Weathered	Surface	5820	5	4	50	35
SW	Transition	5820	5800	10	5	55	47
SW	Fresh	5800	5770	10	5	75	59
SE	Weathered	Surface	5860	5	4	50	37
SE	Transition	5860	5840	10	5	55	47
SE	Fresh	5840	5770	10	5	75	55

Geotechnical issues encountered in the Kombokolo pit in late 2016 led to a pit failure which was thoroughly investigated and is now the basis of the Kombokolo pit slope design and configurations. Additional geotechnical holes drilled in the eastern wall identified a deeper weathering profile of +50 m compared to the initial weathering profile used for the slope designs (Figure 15-8). This showed poor ground conditions and a rating (RMR) of <20.

Figure 15-8 Long Section of Kombokolo Pit Showing Deeper Weathering Profile

Based on the Mining Rock Mass Model and rigorous analyses, four geotechnical domains have been defined as shown in Figure 15-9.

18^th September 2018 Page 233

Kibali Gold Mine NI 43-101 Technical Report

Figure 15-9 Geotechnical Domains for Kombokolo Pit

The pit slope design parameters and configuration were therefore adjusted for the domains. The East domain was revised from 10 m bench height bench to a 5 m bench height due to a deeper weathering profile and the Inter Ramp angle from 43° to 34°. The western domain boundary from 5950 mRL to 5900 mRL was also adjusted from a 10 m bench height to a 5 m bench height (Table 15-11).

This led to an increase in the strip ratio to a larger size pit requiring massive stripping of waste. Pushback designs were used to ensure that the upfront strip ratio was controlled as ore was mined from the higher benches unaffected by the failure, at a minimum rate.

Pit wall monitoring continued in all the active mining pits with the emphasis on the Kombokolo pit and no major movement was registered during the year. The Kombokolo monitoring was improved with the acquisition of a 3D laser mapping monitoring system which is now commissioned, and the data is received and interpreted in near real time.

18^th September 2018 Page 234

Kibali Gold Mine NI 43-101 Technical Report

Table 15-11 Kombokolo Re-Designed Pit Slope Angles

Domain	From   mRL	To   mRL	Bench Height   (m)	Berm Width   (m)	Batter Angle (°)	Inter Ramp   Slope (°)	Weathering
East	Surface	5830	5	4	50	34	Saprolite
	5830	5820	10	5	60	43	Transition
	5820	5810	10	5	75		Fresh
	5810	5730	10	5	80	56	Fresh
North	Surface	5890	5	5	50	43	Saprolite
	5890	5870	5	5	55	47	Transition
	5870	5860	5	5	60		Fresh
	5860	5810	10	5	75		Fresh
	5810	5730	10	5	80	55	Fresh
West	Surface	5920	5	5	50	29	Saprolite
South	Surface	5870	10	5	50	43	Saprolite
	5870	5850	10	5	55	47	Transition
	5850	5840	10	5	60		Fresh
	5840	5790	10	5	75		Fresh
	5790	5730	10	5	80	55	Fresh

For the Gorumbwa pit, four dominant domains were identified and generated (Figure 15-10). Appropriate slope angles were then designed for the various domains, as presented in Table 15-12.

Figure 15-10 Geotechnical Domains for Gorumbwa Pit

18^th September 2018 Page 235

Kibali Gold Mine NI 43-101 Technical Report

Table 15-12 Gorumbwa Recommended Pit Slope Configuration

Domain	Design Section	From mRL	To mRL	Batter Height (m)	Berm Width (m)	Batter Angle (Degrees)	Inter-Ramp Slope  Angle (Degrees)
NE	NE1	Surface	5830	5	5	50	32
NE	NE1	5830	5800	10	4	55
NE	NE1	5800	5750	10	4	65
NE	NE1	5750	5650	10	4	70	50
South	S1	Surface	5820	5	5	50	32
South	S1	5820	5790	10	5	55
South	S1	5790	5750	10	4	60
South	S1	5750	5670	10	4	65
South	S1	5670	5650	10	4	70	48
NW	NW1	Surface	5820	5	5	50	32
NW	NW1	5820	5790	10	4	55
NW	NW1	5790	5750	10	4	60
NW	NW1	5750	5670	10	4	65
NW	NW1	5670	5650	10	4	70	48
Footwall	-	Surface	5820	5	5	50	32
Footwall	-	5820	5650	10	6	55	38

For the Pakaka pit, three dominant domains were identified and generated (Figure 15-11). Appropriate slope angles were then designed for the various domains, as presented in Table 15-13.

Figure 15-11 Geotechnical Domains for Pakaka Pit

18^th September 2018 Page 236

Kibali Gold Mine NI 43-101 Technical Report

Table 15-13 Pakaka Recommended Pit Slope Configuration

Domain	Material	From mRL	To mRL	Batter Height (m)	Batter Angle (°)	Berm Width (m)	Inter-Ramp Slope  Angle (Maximum depth 150 m)
North Domain	Weathered	Surface	5820	5	50	4	33
North Domain	Fresh	5820	5810	10	50	6
North Domain	Fresh	5810	5770	20	60	6
North Domain	Fresh	5770	5730	20	65	6
North Domain	Fresh	5730	5710	20	70	6	52
South Domain	Weathered	Surface	5840	5	50	4	33
South Domain	Fresh	5840	5830	10	50	6
South Domain	Fresh	5830	5810	20	55	6
South Domain	Fresh	5810	5770	20	60	6
South Domain	Fresh	5770	5750	20	65	6	50
Footwall Domain	Weathered	Surface	5830	5	50		33
Footwall Domain	Fresh	5830	5750	20	55

For the Pamao pit, four dominant domains were identified and generated. Appropriate slope angles were then designed for the various domains as presented in Table 15-14.

Table 15-14 Pamao Recommended Pit Slope Configuration

Domain	From mRL	To mRL	Bench Height (m)	Berm Width (m)	Batter Angles (°)	Inter Ramp Slope Angle (°)	Weathering
FW	Surface	5840	5	5	45	38	Saprolite/Oxide
	5840	5720	10	5	80	55	Fresh
West	Surface	5800	5	5	50	34	Saprolite/Oxide
	5800	Bottom	10	5	80	55	Fresh
HW	Surface	5840	5	5	55	42	Saprolite/Oxide
	5840	Bottom	10	5	80	56	Fresh
East	Surface	5810	5	5	50	36	Saprolite/Oxide
	5810	Bottom	10	5	80	55	Fresh

Pit Dewatering

The dewatering strategy implemented at the Pakaka pit by use of dewatering boreholes along the pit perimeter combined with in-pit sumps led to the successful completion of mining the PB1 pit. Electric pump installations were designed to keep the pit dry until the start of mining of the PB2 pit in 2020, and these installations will be ready by Q2 2019. A diversion trench has also been designed as part of the strategy to manage runoff from the south-eastern catchment to prevent flooding in the pit after a 1:50 year rainfall event. Some dewatering boreholes were also drilled in advance to depress the water table ahead of scheduled mining.

There was no strategy change in Kombokolo as the perimeter boreholes and in-pit sump successfully kept the pit dry for mining operations.

Borehole pumping along the western perimeter of the Sessenge pit has also successfully kept the water table at a lower level from the mining faces (Figure 15-12). A diversion trench was designed for the NE side of the pit to capture all runoff from the KCD PB3 hill.

18^th September 2018 Page 237

Kibali Gold Mine NI 43-101 Technical Report

Figure 15-12 Layout of Dewatering Boreholes in Sessenge Pit

The old shaft in Gorumbwa pit was pumping at a rate of 39 l/s to dry the voids left behind by the old underground mine. Dewatering was down to 140 m below surface, allowing Phase 1 of scanning of the underground voids to take place (Figure 15-13). Large capacity electrical pumps were sourced to improve the pumping rate to 60 l/s; this is in preparation for Gorumbwa pit mining in Q1 2019. The pump installation is being lowered 200 m down the shaft to dry up the remaining 40 m and achieve dewatering below the pit bottom by end of Q2 2018. This will keep the water below the pit until the shaft is decommissioned and replaced by conventional boreholes at the start of mining activities.

The KCD open pit was dewatered through a combination of boreholes and sump pumping. As underground workings developed and the pit deepened the majority of dewatering wells ran dry due to declining water levels, and there are now no active dewatering wells at KCD.

The Pamao pit dewatering system will be designed and commissioned prior to when the operations begin in 2021.

18^th September 2018 Page 238

Kibali Gold Mine NI 43-101 Technical Report

Figure 15-13 Section of Gorumbwa Drawdown in Old Voids

Underground Mine

Geotechnical Aspects

SRK Consulting (SRK) completed the original geotechnical assessment as part of the feasibility study for the Kibali Underground Project in 2011. At the request of Kibali Goldmines, this work was externally reviewed by KSCA Geomechanics Pty Limited, where the main focus of the review was to identify any points or sections within the SRK study report that it was felt should be checked, followed up, or re-examined. This external review was completed in 2012, with several recommendations and suggestions identified as opportunities to improve on the overall geotechnical understanding of the Kibali underground environment.

During and after the feasibility study into the detailed engineering and early implementation phase, mining rock mass and structural models, stress measurements, and empirical and numerical analyses were completed. The completion of this work gave a significantly improved understanding of the expected geotechnical environment. In line with this compilation of work, Kibali Goldmines agreed that the Project would adopt a good-practice approach to the operation of the underground mine. This resulted in the mine adopting the Western Australian Mines Safety and Inspection Regulations (1995). The mine achieved compliance with Regulation 10.28, thereby guaranteeing adequate geotechnical considerations when planning, designing, and operating the mine. This approach continues to the present.

Since the completion of the feasibility study in 2011 by SRK Consulting, underground mining geotechnical assessments for Kibali have been undertaken through several different external consultancy companies, namely Dempers & Seymour, Coffey Mining, Beck Engineering, KSCA Geomechanics Pty Limited and the Western Australian School of Mines. Combined, these consultancy companies have covered and completed different geotechnical aspects of work – this has resulted in a significant amount of work, including the following:

18^th September 2018 Page 239

Kibali Gold Mine NI 43-101 Technical Report

☐ Development and construction of mining rock mass and structural models between 2012 and 2017. These models have incorporated raw geotechnical data comprised of rock mass and structural logs of exploration, geotechnical drill holes and underground mapping - the current data includes rock mass logging and structural measurements from drilling, in addition to structural measurements from underground mapping .

☐ 3D mine-wide numerical modelling, completed to understand the resultant mining induced effects from proposed mining sequences in terms of stress effects, damage and seismic potential in relation to the different ore zones and surrounding infrastructure.

☐ In situ pre-mining stress measurements

☐ Establishment of a Stope Performance Database to allow for the collection of relevant geotechnical stope design parameters, which then enable a comparison to be made between predicted and actual stope behaviour for each stope.

☐ Establishing the Kibali Stope Performance in relation to the Stability Graphs.

With the underground stope production starting in December 2014 (and the subsequent voids backfilled with paste), assessing the geotechnical aspects of Kibali has continued over the recent years. It is anticipated that this will continue as experience is gained from understanding how the rock mass responds overtime to the mining induced effects from stope production. This work will be particularly pertinent if there are any future changes to the geology block model that then requires modifications to stope shapes and the subsequent extraction sequence.

As with previous Kibali Ore Reserve Updates, numerical modelling has been used to confirm that new stope shapes and extraction sequence are ‘fit for purpose’ as part of the Ore Reserve process. The modelling is considered to be an important means in providing a tool to quantify stability and performance of stope extraction and identify the stope extraction optimisation potential. The numerical modelling is also a method from which risk can be gauged, understood, and mitigated.

This approach has continued for the 2017 Ore Reserve Update, with the work again being undertaken externally by Beck Engineering. The numerical modelling started in November 2017, with the results due to be completed during March 2018. This modelling will consist of a base case simulation and an additional simulation if modifications to the sequence are required in order to manage and mitigate any resultant adverse mining induced effects. The learning’s gained from previous Ore Reserve numerical modelling will be included in the 2017 updating process when the initial base case extraction sequence is established.

Hydrogeology

Pumping installations in the sections of the mine accessed from the decline are in operation. A permanent pumping station in the shaft has been commissioned and provides pumping of all water from the underground mine.

Hydrogeological modelling and monitoring at KCD underground is managed by a hydrogeological team on site. SRK provides support and review on hydrogeological aspects. This section is largely based on a site visit by SRK in December 2017.

18^th September 2018 Page 240

Kibali Gold Mine NI 43-101 Technical Report

A summary of the dewatering history at KCD is provided below for context:

☐ The KCD open pit was dewatered through a combination of boreholes and sump pumping.

☐ As underground workings developed, and the pit deepened the majority of dewatering wells ran dry due to declining water levels, and there are now no active dewatering wells at KCD.

☐ The majority of exploration boreholes drilled at KCD were not grouted. Intersection of exploration boreholes has been the greatest dewatering challenge at KCD to date. Not all exploration holes were accurately surveyed, and hole deviation is often severe, presenting challenges to planning for hole intersection underground.

☐ Permeable structures associated with dolerite dykes and ironstone formations has also presented an inflow risk, especially at the base of the shaft, prior to commissioning of the clean water pumping station.

☐ The objective of the dewatering programme at KCD is full depressurisation of deep and shallow aquifers. This is being achieved by allowing grade control and exploration holes to flow when intersected (wherever possible).

☐ Drain hole drilling has been undertaken during the development of the declines and from the base of the shaft to pre-drain permeable structures, avoiding uncontrolled inflows to development, stopes, and the southern haulage roadway. The drain hole drilling has also increased the clean water make and reduced dirty water make, which has mitigated the impact of delays in commissioning the dirty water treatment system at the base of shaft pumping station.

☐ Inflows from the open pits have caused temporary flooding of B zone and has delayed depressurisation of aquifers above C zone.

☐ Dewatering flow rates are broadly in line with those predicted during the Feasibility Study, with groundwater inflows being slightly lower and inflows from the open pit being higher than predicted.

Service water flows are typically between 25 to 30 l/s; Groundwater inflows are typically in the range of 60 to 70 l/s. Inflows associated with the flooding of KCD South pit range from 0 to >30 l/s (the precise upper limit is unknown as it results in flooding of B zone).

18^th September 2018 Page 241

Kibali Gold Mine NI 43-101 Technical Report

15.4 Dilution and Mining Recovery

Open Pits

A detailed dilution exercise was undertaken for the KCD Pushback 3 deposit to determine the expected dilution on the main lodes. This was based on the fleet type to be used and the mining method employed.

Dilution, when mining around voids has been included at 13%, while dilution at 9.6% has been included when mining the larger lodes. An average of 11.3% dilution has been estimated for the eight upper flitches of the pit. The average dilution can be expected to decrease when the void areas are mined out and mining takes place in normal mining conditions without voids. A 10% dilution allowance has therefore been adopted for the entire KCD deposit.

Ore loss has been estimated at 8% for the larger void areas in the KCD pit; the ore is expected to be able to be retrieved from lower parts of the voids and so can be considered a delayed recovery rather than total ore loss. A global 3% factor has been used for ore losses in the estimation of open pit Ore Reserves.

The QP considers that the dilution and loss factors are reasonable assumptions for the estimation of the Ore Reserves.

Underground Mine

The underground mining methods being used are variants of long hole open stoping with cemented paste, namely:

☐ Primary / Secondary long hole open stoping

☐ Advancing face long hole open stoping

☐ Longitudinal open stoping

Three forms of dilution have been included; planned dilution, unplanned dilution, and paste fill dilution.

Planned dilution is either interburden between lodes where multiple lodes are included in a stope shape, footwall waste included to enable mucking of ore, or hangingwall material included to create a supportable hangingwall. Planned dilution is added at actual block model grades.

Unplanned dilution is added as a percentage of stope tonnes. Unplanned dilution is rock dilution outside of the designed mining shapes on the footwall or hangingwall of the stopes. Unplanned dilution is added at a gold grade of 0 g/t.

Paste dilution is added where a stope has a paste fill exposure. Paste dilution exposures have been estimated as 1% per paste fill exposure. This is essentially a 0.3 m thickness of paste on any exposure. Hence some secondary stopes with three paste exposure walls will have 3% paste

18^th September 2018 Page 242

Kibali Gold Mine NI 43-101 Technical Report

dilution. It should be noted that overbreak into adjacent stopes (“ore gain”) can be a predictor of paste dilution in adjacent stopes.

Total Unplanned dilution - Unplanned rock dilution and paste dilution have been combined and applied as a combined percentage.

Stopes mined in 2017 showed an “ore gain” of 4% or approximately 0.4 m thickness. “Ore gain” is material above the cut-off grade that was planned to be mined later in an adjacent stope, but due to overbreak has been mined early. It is assumed that continuous improvement of drill / blast processes will lead to a reduction in this overbreak. “Ore gain” has not been included in the estimation of Ore Reserves.

The Kibali site technical team routinely reconciles actual stope and development production against planned performance. The QP has reviewed the reconciliations and considers that the process is producing a reasonable analysis of actual performance.

A summary of the 2017 results is shown in Table 15-15.

Table 15-15 Summary of 2017 Mining Recovery and Unplanned Dilution

Stope Name	Primary/Secondary	Recovery                  Actual Ounces	Dilution
C_630_666_XC8_MS20	P	82%	9%
C_560_490_XC08_MS#14/MS15/MAS16	P	97%	6%
C_490_XC14_455_MS	P	91%	1%
C_595_630_XC_8_MS#21	P	90%	7%
Subtotal Primary		93%	4%
B1_580_555_XC7-LS#13	S	92%	9%
B1_605_XC3_58-_LS#08	S	94%	6%
B1_580_XC7_555_LS-10	S	94%	6%
C_630_665_XC3-MS-11	S	80%	7%
B1_605_630_XC1 (LS#20)	S	76%	4%
C_560_525_XC7	S	83%	3%
B1_580_605_XC_5 (LS#17)	S	83%	3%
Subtotal Secondary		87%	5%
Total 2017		89%	5%

In 2017, ore loss and dilution have been higher than expected in the year end 2016 Ore Reserve estimate. This is due to the following factors:

☐ The mine is early in its production phase and is in a learning process about the drill / blast and geotechnical performance of the stopes.

☐ A significant portion of stopes mined to date are in the “B” area 5105 lode. The 5105 is flatter than 5101 and 5102 lodes and comprises only approximately 2% of the remaining Ore Reserve.

☐ Initial stopes were filled with paste fill made using Portland Pozzolana cement (PPC) binder and a faster strength deterioration was experienced than expected. Subsequent to this, slag binder has been used which shows better strength results and it is anticipated that this will lead to lower backfill dilution.

Dilution parameters used for the year end 2017 Ore Reserve estimate have been summarised in Table 15-16.

18^th September 2018 Page 243

Kibali Gold Mine NI 43-101 Technical Report

Table 15-16 Summary of 2017 Underground Ore Reserve Estimate Dilution Parameters

Stoping Type	Sequence Configuration	Rock Exposures			Unplanned Rock Dilution	Fill Exposures	Paste Dilution	Total Unplanned + Paste Dilution
		Foot Wall	Hanging Wall	Back
Transverse Primary Stope	Hanging Wall		1		3%		0%	3%
	Not Hanging Wall				0%	1	1%	1%
Transverse Secondary Stope	Hanging Wall		1		3%	2	2%	5%
	Not Hanging Wall				0%	3	3%	3%
Transverse Advancing Face Stope	Hanging Wall		1		3%	2	2%	5%
	Not Hanging Wall				0%	2	2%	2%
Longitudinal Stope	Hanging Wall	1	1		12%	1	1%	13%
	Not Hanging Wall	1			3%	1	1%	4%

An ore loss of 3% (97% mining recovery) has been applied for most stopes. A higher ore loss of 10% has been applied to some secondary stopes in 5101/5102. In the 5101/5102 transverse stoping area, a modified sequence has been adopted. Previously it was planned to mine the lower (5101) stopes first and advance upwards into the 5102 stopes. The additional ore loss in these stopes is 30 koz Au. The modified sequence involves mining some of the 5102 primary stopes in 2017/2018, before the 5101 stopes below. This sequence is expected to create mining induced stress issues in some of the 5101/5102 secondary stopes. The modelling to quantify this issue is currently underway.

In comparison with the year end 2016 Ore Reserve estimate, the following was noted:

☐ Planned rock dilution is assessed to be the same as the previous estimate.

☐ Unplanned rock dilution has been increased. Previously, transverse stopes with a hanging wall contact had 2.7% and transverse stopes with no hanging wall contact had 1% dilution. Longitudinal stopes had 6.7% dilution.

☐ Paste dilution has increased from 1%, in total, to 1% per paste exposure.

☐ Ore loss is the same as the 2016 estimate at 3% (97% mining recovery).

15.5 Economic Parameters

Open Pits

The cut-off grades have been estimated for each material type for all six reserve pits included in the 2017 Ore Reserve estimate. These are based on a gold price of $1,000/oz and $1,100/oz for the KCD PB3 pit and include dilution, royalties, processing cost and recoveries, general and administration cost, and ore mining costs.

Gold Price and Royalties

With the exception of the KCD pit, which assumes a gold price of $1,100/oz, all the other reserve pits are based on a gold price of $1,000/oz. This is in line with Randgold corporate guidelines, which consider the long-term gold price forecasts. Gold price sensitivities were prepared for all the pits, the decision on a higher price for the KCD pit is discussed in detail under Section 15.6.

18^th September 2018 Page 244

Kibali Gold Mine NI 43-101 Technical Report

Royalties payable to the DRC government remained unchanged from the year end 2016 estimate. A total royalty of 3.5% of gold revenue inclusive of 1% shipment fees was used for the year end 2017 estimate.

Processing Costs

Processing costs for the year were reviewed and the numbers were found to be in line with the 2016 LOM projections. The 2016 costs of $16.10/t and $14.34/t for sulphide ores and Oxide/transition ores, respectively, were therefore maintained for the 2017 Ore Reserve estimate.

General and Administration Cost

The G&A cost for Kibali mine was reviewed based on LOM expectations and actuals for the year end 2017. An upward adjustment of 9% ($7.40/t from $6.80/t) was noted and this was subsequently applied in the 2017 Ore Reserve estimation.

Mining Costs

The mining costs used for the 2017 pit optimisations were derived from the KMS 2017 BUP and Long-Term Review (LTR) pricing for the Kibali open pit operations.

Mining Cost Adjustment Factors (MCAF’s) were generated from various bench by bench waste mining costs received for all the deposits. The waste mining cost is inclusive of fuel cost, drill and blast cost per bench, pre-split cost, explosive cost per tonne, mining departmental cost, pit dewatering, rehabilitation cost, and contractor fixed costs.

The MCAF’s were then imported into their respective block models and assigned to the corresponding benches in Surpac software for the creation of economic block models.

The mining cost for the KCD pit was adjusted upwards to allow for the higher cost of local mining contractors using smaller fleets (ADTs) during the pit start up to cater for the undulating topography and narrow benches.

The Ore Reserves are based on a marginal cut-off grade. Mineral Resources contained within the final pit designs were evaluated against these cut-off grades to produce the Open Pit Proved and Probable Ore Reserves

Cut-off grade sensitivities were trialled by adjusting the gold price (Table 15-17 and Table 15-16).

18^th September 2018 Page 245

Kibali Gold Mine NI 43-101 Technical Report

Table 15-17 KCD, Kombokolo, Pakaka Open Pits - Marginal and Full Grade Ore Cut-Off Grade for Different Material Types

	Deposit			KCD				Kombokolo				Pakaka
	Material Type	Unit	Constant	Oxide	Tans	Fresh	Total	Oxide	Tans	Fresh	Total	Oxide	Tans	Fresh
Mining	Waste cost (per tonne mined)	$/t		2.92	2.97	3.09	2.99	2.57	2.69	2.87	2.71	2.72	2.80	2.88
	Extra Ore Cost (per ore tonne)-GC+Core-rehandle+Overhaul	$/t		1.27	1.27	1.27	1.27	1.19	1.19	1.19	1.19	1.38	1.38	1.38
	GC Only (per tonne mined)	$/t		0.75	0.75	0.75	0.75	0.75	0.75	0.75	0.75	0.75	0.75	0.75
	Dilution	%	10	10
	Ore Loss	%	3	3
Process	Haulage cost per ore ton	$/t	2.50									1.05	1.05	1.05
	Process Cost (per ore tonne mined)	$/t	14.93	14.34	14.34	16.10	14.93	14.34	14.34	14.34	14.34	14.34	14.34	16.10
	Process Recovery	%		90.1	90.1	86.1	88.8	85.0				88.7	81.3	80.2
	Plant Throughput	Mtpa	7.2	7.20
G&A	General Admin (per ore tonne mined)	$/t	7.4	7.40
Revenue	Gold Price (Reserve)	$/oz		1,100				1,000
	Gold Price	$/gm	31.10348	35.37				32.15
	Gold Royalites	$/oz	3.50%	38.50				35.00
	Net Gold Price	$/oz		1061.50				965.00
	Net Gold Price	$/gm		34.12				31.00
	Resource Gold Price	$/oz	1,500	1,500
	Total Process Cost (per ore tonne mined)	$/t		14.34	14.34	16.10	14.93	14.34	14.34	14.34	14.34	15.39	15.39	17.15
	Total Mining Cost (per ore tonne mined)	$/t		16.15	16.40	17.01	16.52	27.90	29.12	31.02	29.35	15.33	15.75	16.19
	Marginal Cut-off Grade	g/t		0.82	0.82	0.93	0.86	0.87	0.87	0.87	0.87	0.88	0.96	1.04
	Strip Ratio						4.1				9.4
	FGO Cut-off Grade	g/t		1.36	1.36	1.52	1.41	1.60	1.65	1.72	1.66	1.39	1.53	1.64

18^th September 2018 Page 246

Kibali Gold Mine NI 43-101 Technical Report

Table 15-18 Pamao, Sessenge, Gorumbwa Open Pits - Marginal and Full Grade Ore Cut-Off Grade for Different Material Types

	Deposit				Pamao				Sessenge				Gorumbwa
	Material Type	Unit	Constant	Total	Oxide	Tans	Fresh	Total	Oxide	Tans	Fresh	Total	Oxide	Tans	Fresh	Total
Mining	Waste cost (per tonne mined)	$/t		2.80	2.85	2.88	2.95	2.89	2.62	2.68	2.80	2.70	2.92	3.14	3.24	3.10
	Extra Ore Cost (per ore tonne)-GC+Core- rehandle+Overhaul	$/t		1.38	1.31	1.31	1.31	1.31	1.24	1.24	1.24	1.24	1.28	1.28	1.28	1.28
	GC Only (per tonne mined)	$/t		0.75	0.75	0.75	0.75	0.75	0.75	0.75	0.75	0.75	0.75	0.75	0.75	0.75
	Dilution	%	10	10
	Ore Loss	%	3	3
Process	Haulage cost per ore ton	$/t	2.50	1.05	1.05	1.05	1.05	1.05
	Process Cost (per ore tonne mined)	$/t	14.93	14.34	14.34	14.34	16.10	14.34	14.34	14.34	16.10	14.34	14.34	14.34	14.34	14.34
	Process Recovery	%		83.4	90.9	85.0	83.5	86.5	93.1	88.6	67.0	82.9	90.0	90.0	90.0	90.0
	Plant Throughput	Mtpa	7.2	85.0
G&A	General Admin (per ore tonne mined)	$/t	7.4	7.20
Revenue	Gold Price (Reserve)	$/oz		7.40
	Gold Price	$/gm	31.10348	1,000
	Gold Royalites	$/oz	3.50%	32.15
	Net Gold Price	$/oz		35.00
	Net Gold Price	$/gm		965.00
	Resource Gold Price	$/oz	1,500	31.00
	Total Process Cost (per ore tonne mined)	$/t		15.39	15.39	15.39	17.15	15.39	14.34	14.34	16.10	14.34	14.34	14.34	14.34	14.34
	Total Mining Cost (per ore tonne mined)	$/t		15.76	8.79	8.88	9.05	8.91	11.98	12.23	12.71	12.31	33.38	35.82	36.94	35.38
	Marginal Cut-off Grade	g/t		0.93	0.85	0.91	1.00	0.90	0.88	0.92	1.31	0.98	0.82	0.82	0.82	0.82
	Strip Ratio			4.1				1.6				3.1				10.0
	FGO Cut-off Grade	g/t		1.49	1.12	1.20	1.30	1.18	1.28	1.36	1.92	1.46	1.97	2.06	2.10	2.05

18^th September 2018 Page 247

Kibali Gold Mine NI 43-101 Technical Report

The QP responsible for the Ore Reserve estimate considers that the process used is appropriate for the estimation of open pit Ore Reserves at Kibali Mine.

Underground Mine

The stope marginal cut-off grade used for the estimation of the 31st December 2017 underground mine Ore Reserves is the diluted cut-off grade. The cost of processing and site administration cost has been included in the calculation of the marginal cut-off grade.

Table 15-19 shows the cut-off grade calculation for the underground mine.

Table 15-19 Kibali Underground Mine – Cut-Off Grade Calculation

Revenue Parameters
Description	Units	Value	Comment
Gold Price	$/oz	1000	2017 Reserve Price
Process Plant Gold Recovery	%	85	Tham, Nic (2017)
Royalty	$/oz	35	Naude (2017)
Cost Parameters
Mine Production	$/t	41	2018 and 2019 UG Cost for Draft Budget
Capital	$/t	0	Excluded - OPEX Only Assessment
Backfill	$/t	0	Included in Mine Production Cost
Processing	$/t	16.1	LOM Process Cost (Naude, 2017)
Site G&A	$/t	7.4	LOM G&A (Naude, 2017)
Total Unit Cash Costs	$/t	64.5
Mining Cut Off Grade	g/t	2.5

For the majority of the Kibali underground Ore Reserve a cut-off grade of 2.5 g/t Au has been used. In the 9105 lode, the cut-off grade has been increased to 2.8 g/t Au. These cut-off grades were applied to stope panels after dilution and ore loss had been accounted for in the panel.

Lode 9105 contains approximately 15% of the contained metal of the Ore Reserve, although with a lower average grade. A grade / tonnage sensitivity evaluation was undertaken on 9105 and it was found that at an elevated cut-off grade of 2.8 g/t Au, the stoping area remains continuous, and as a result the average grade of Lode 9105 has been increased by approximately 0.3 g/t Au.

The Ore Reserve QP considers that the process used to calculate the cut-off grade is appropriate for the Kibali underground mine.

15.6 Open Pit Optimisations

Data checks on block models received from the Mineral Resource Department were conducted. This included checks for missing cells, absent values, density checks, grade errors and correctly assigned weathering profiles. All models received had waste blocks built into them.

Economic models were generated from the Resource block models with the inclusion of the MCAF’s for each of the six target deposits. Approved geotechnical slope domains and angles based on rock characteristics and behaviour were also assigned to the block models before converting them to block models suitable for optimisation. These were then imported into Whittle Four-X for the pit optimisation exercise.

18^th September 2018 Page 248

Kibali Gold Mine NI 43-101 Technical Report

The initial optimisation run considered the Measured and Indicated Resources with Inferred Mineral Resources excluded. These were run with a gold price of $1,000/oz for Ore Reserves.

A second set of optimisations was conducted with the inclusion of Inferred Mineral Resources. These optimisations were used to quantify the Inferred portions of the deposits, determine the impact on the mine plan, and to provide direction to the Geology and exploration department for possible targets for drilling and resource conversion.

The advanced grade control and infill drilling campaign, started in 2017, resulted in major changes to several of the pits, notably Kombokolo, Sessenge, KCD, and Pamao. Pakaka and Gorumbwa pits remained fairly constant with minimal changes. Table 15-20 shows the comparison between the pit shells for Sessenge from the 2016 Whittle run to the 2017 Whittle run and

Table 15-21 shows the comparison between the pit shells for Pamao from the 2016 Whittle run to the 2017 Whittle run.

For the KCD deposit, which is the main deposit and is also extracted by underground methods, the optimisation was run by constraining the block model to the underground-open pit interface at 5680 mRL. The block model imported into Whittle was therefore constrained to this elevation. The optimal pit shell is the $1,100/oz shell, which delivers higher ounces and no downside risk at a lower gold price.

No other physical surface infrastructures constraints were applied for any of the orebodies in the Ore Reserves.

Table 15-20 Comparison of Whittle Results for Sessenge Pit with 2016 Results

$1,000 Shell	2016 Whittle Run	2017 Whittle Run	Diff	% Diff
Probable
Ore Tonnes (t)	5,476,203	3,617,820	-1,858,383	-34%
Ore Grade (g/t)	1.46	1.73	0.91	19%
Total Indicated Ounces	256,219	201,792	-54,427	-21%
Inferred
Inferred Tonnes (t)	2,099,218	494,353	-1,604,865	-76%
Inferred Grade (g/t)	1.92	1.73	2	-10%
Total Inferred Ounces	129,888	27,509	-102,379	-79%
Total
Total Ore Tonnes	7,575,421	4,112,173	-3,463,248	-46%
Grade	1.59	1.73	1	9%
Total Ounces	386107	229301	-156,806	-41%
Total Waste Tonnes	13,456,297	8,746,888	-4,709,409	-35%
Strip Ratio with Inferred Ore	1.8	2.1	0.35	20%
Strip Ratio without Inferred Ore	2.8	2.6	-0.29	-10%
% of Inferred Tonnes	28%	12%
% of Inferred Ounces	34%	12%

18^th September 2018 Page 249

Kibali Gold Mine NI 43-101 Technical Report

Table 15-21 Comparison of Whittle Results for Pamao Pit with 2016 Results

$1,000 Shell	2016 Whittle Run	2017 Whittle Run	Diff	% Diff
Probable
Ore Tonnes (t)	4,290,957	4,047,135	-243,822	-6%
Ore Grade (g/t)	1.96	2.41	- 5.56	23%
Total Indicated Ounces	269,913	313,468	43,555	16%
Inferred
Inferred Tonnes (t)	—	103,160	103,160	0%
Inferred Grade (g/t)	—	2.01	2.01	0%
Total Inferred Ounces	—	6665	6,665	0%
Total
Total Ore Tonnes	4,290,957	4,150,295	-140,662	-3%
Grade	1.96	2.4	-11.10	23%
Total Ounces	269,913	320,132	50,219	19%
Total Waste Tonnes	13,212,235	18,677,447	5,465,212	41%
Strip Ratio with Inferred Ore	3.1	4.5	1.4	46%
Strip Ratio without Inferred Ore	—	4.6	4.6
% of Inferred Tonnes	0%	2%
% of Inferred Ounces	0%	2%

Sensitivity Analysis

An initial optimisation was run on the standard $1,000/oz Reserve gold price. Gold price sensitivities were then run for gold prices of $400/oz to $2,000/oz at an increment of $100/oz to produce a set of nested pits shells (Table 15-22 to Table 15-25 and Figure 15-14).

18^th September 2018 Page 250

Kibali Gold Mine NI 43-101 Technical Report

Table 15-22 Sessenge Gold Price Sensitivities

Pit Size	Cash Flow	Ore	Grade	Waste	Mining Cost	Process Cost	Royalties	Mining Cost	S/R	Ounces Mined	Process Recovery	Gold Produced	Cash cost
$/oz	$	Mt	g/t	Mt	$ M	$ M	$ M	$/t	t:t	koz	%	oz	$/oz
400	52	1.2	2.78	1.0	-1	-28	-3	2.57	0.8	106	83.2	88	414
500	73	1.9	2.71	2.4	-11	-44	-5	2.58	1.3	163	82.1	134	450
600	87	2.5	2.62	4.2	-17	-59	-6	2.58	1.7	207	81.7	169	485
700	97	3.0	2.57	6.7	-25	-72	-7	2.59	2.2	248	81.3	202	517
800	101	3.4	2.51	8.2	-30	-81	-8	2.59	2.4	271	81.2	220	539
900	104	3.8	2.45	10	-36	-91	-8	2.59	2.7	296	81	240	565
1000	106	4.2	2.4	13	-44	-101	-9	2.6	3.1	320	80.9	259	592
1100	105	4.5	2.35	14	-49	-108	-10	2.61	3.2	336	80.8	272	614
1200	103	4.7	2.31	16	-55	-114	-10	2.61	3.5	349	80.7	282	635
1300	90	5.7	2.22	28	-89	-140	-12	2.63	4.9	409	80.5	329	728
1400	82	6.2	2.2	34	-106	-151	-12	2.64	5.5	437	80.4	351	767
1500	81	6.2	2.19	35	-107	-152	-12	2.64	5.6	439	80.4	353	770
1600	78	6.3	2.18	36	-112	-155	-13	2.64	5.7	445	80.4	358	782
1700	77	6.4	2.18	37	-113	-156	-13	2.64	5.7	446	80.4	359	785
1800	58	6.9	2.15	47	-142	-168	-13	2.64	6.8	474	80.3	381	848
1900	54	6.9	2.14	49	-147	-170	-13	2.65	7	479	80.3	385	860
2000	53	7.0	2.14	49	-149	-170	-13	2.65	7.1	480	80.3	385	864

18^th September 2018 Page 251

Kibali Gold Mine NI 43-101 Technical Report

Table 15-23 Kombokolo Gold Price Sensitivities

Pit Size	Cash Flow	Ore	Grade	Waste	Mining Cost	Process Cost	Royalties	Mining Cost	S/R	Ounces Mined	Process Recovery	Gold Produced	Cash cost
$/oz	$	Mt	g/t	Mt	$ M	$ M	$ M	$/t	t:t	koz	%	oz	$/oz
400	24	0.3	4.12	1.3	-4.7	-7.6	-1.3	2.87	3.9	44	85	37	364
500	37	0.6	4.01	3.5	-12	-13	-2.3	2.88	5.9	76	85	65	425
600	41	0.7	3.93	4.6	-15	-16	-2.6	2.88	6.6	89	85	75	452
700	44	0.8	3.69	5.3	-18	-19	-2.9	2.88	6.4	98	85	83	474
800	48	1.1	3.49	7.7	-25	-24	-3.6	2.88	7.3	119	85	101	525
900	48	1.1	3.41	8.0	-26	-25	-3.6	2.88	7.2	122	85	104	533
1000	49	1.2	3.32	9.1	-30	-28	-3.9	2.88	7.5	130	85	110	557
1100	48	1.3	3.19	10	-32	-30	-4.0	2.87	7.5	135	85	115	579
1200	47	1.4	3.07	11	-36	-33	-4.2	2.86	7.7	142	85	121	606
1300	47	1.5	3.05	11	-36	-33	-4.2	2.86	7.8	143	85	121	610
1400	40	1.8	2.82	17	-54	-42	-4.9	2.85	9.4	165	85	140	717
1500	38	1.9	2.78	18	-57	-43	-5.0	2.85	9.6	169	85	144	735
1600	37	1.9	2.76	19	-60	-44	-5.1	2.85	9.8	172	85	146	749
1700	33	2.1	2.7	21	-67	-47	-5.3	2.84	10.3	179	85	152	784
1800	31	2.1	2.68	22	-69	-48	-5.4	2.84	10.6	181	85	154	798
1900	24	2.3	2.6	26	-80	-52	-5.7	2.84	11.3	191	85	162	850
2000	24	2.3	2.6	26	-80	-52	-5.7	2.84	11.3	191	85	163	851

18^th September 2018 Page 252

Kibali Gold Mine NI 43-101 Technical Report

Table 15-24 Pamao Gold Price Sensitivities

Pit Size	Cash  Flow	Ore	Grade	Waste	Mining  Cost	Process Cost	Royalties	Mining  Cost	S/R	Ounces  Mined	Process Recovery	Gold Produced	Cash cost
$/oz	$	Mt	g/t	Mt	$ M	$ M	$ M	$/t	t:t	koz	%	oz	$/oz
400	1.4	0.03	2.48	0.01	-0.1	-0.8	-0.1	4.78	0.2	2.6	86.5	2.6	374
500	11	0.32	2.34	0.30	-1.8	-7.5	-0.8	4.15	0.9	24	86.5	24	426
600	27	1.0	2.10	1.6	-7.7	-25	-2.5	3.87	1.5	70	86.5	70	493
700	38	1.7	1.99	2.7	-13	-40	-3.8	3.83	1.6	108	86.5	108	525
800	51	3.0	1.85	6.1	-27	-71	-6.2	3.68	2.0	178	86.5	178	582
900	55	3.6	1.78	7.4	-32	-86	-7.3	3.68	2.0	207	86.5	207	604
1000	56	4.1	1.74	8.7	-38	-98	-8.0	3.65	2.1	229	86.5	229	625
1100	54	5.4	1.60	11	-48	-122	-10	3.65	2.1	275	86.5	275	652
1200	38	8.4	1.56	30	-110	-179	-15	3.38	3.6	422	86.5	422	720
1300	25	10.5	1.49	38	-141	-213	-18	3.37	3.6	507	86.5	507	733
1400	22	11.2	1.46	40	-146	-218	-18	3.38	3.5	527	86.5	527	726
1500	14	12.1	1.43	43	-159	-227	-19	3.37	3.6	555	86.5	555	729
1600	-5	13.5	1.39	52	-187	-239	-21	3.34	3.8	605	86.5	605	739
1700	-13	14.3	1.37	55	-199	-244	-22	3.34	3.9	627	86.5	627	741
1800	-17	14.9	1.34	56	-203	-246	-22	3.35	3.8	641	86.5	641	736
1900	-21	15.4	1.32	57	-209	-248	-23	3.35	3.7	652	86.5	652	735
2000	-28	16.0	1.30	60	-217	-251	-23	3.35	3.7	668	86.5	668	736

18^th September 2018 Page 253

Kibali Gold Mine NI 43-101 Technical Report

Table 15-25 KCD Gold Price Sensitivities

Pit Size	Cash Flow	Ore	Grade	Waste	Mining Cost	Process Cost	Royalties	Mining Cost	S/R	Ounces Mined	Process Recovery	Gold Produced	Cash cost
$/oz	$	Mt	g/t	Mt	$ M	$ M	$ M	$/t	t:t	koz	%	oz	$/oz
800	95	3.2	2.56	11	-43	-81	-8	2.93	3.6	263	86.5%	227	582
900	101	3.7	2.44	13	-49	-89	-9	2.94	3.5	287	86.5%	248	591
1000	107	4.7	2.24	18	-66	-111	-10	2.94	3.8	340	86.5%	294	637
1100	121	6.3	2.09	25	-92	-142	-13	2.96	3.9	426	86.5%	369	671
1200	123	7.0	2.00	28	-102	-151	-14	2.96	3.9	452	86.4%	391	685
1300	130	7.5	1.97	30	-111	-157	-14	2.96	4.0	478	86.4%	413	685
1400	131	7.9	1.92	31	-115	-159	-15	2.96	3.9	487	86.1%	420	688
1500	132	8.2	1.89	32	-120	-162	-15	2.96	3.9	499	86.1%	430	692

18^th September 2018 Page 254

Kibali Gold Mine NI 43-101 Technical Report

Figure 15-14 KCD Sensitivity Analysis

Analysis of the sensitivities for the KCD pit showed that, when compared to the $1,000/oz pit, the $1,100/oz pit provides a deeper pit, containing an extra 1.6 Mt at 1.64 g/t Au containing 86 koz at a similar strip ratio and with no need for a pushback. Most of the extra ore is deep-seated with 954 kt at 2.17 g/t Au containing 67 koz of gold below the $1,000/oz pit together with a further 687 kt at 0.89 g/t Au containing 19 koz of gold as a result of the lower cut-off grade.

It should be noted that in order to convert this deeper material, which is predominantly Inferred Mineral Resources (80% of the difference between the $1,000/oz and $1,100/oz pit shells), into Ore Reserves, closer spaced grade control infill drilling will be required to increase the geological confidence level when mining progresses.

Pit Selection

Apart from the KCD pit, which was based on a $1,100/oz pit shell, all the other reserve pits were designed on a $1,000/oz price pit shell selected following an analysis of the pit size against value and gold price.

The analysis prior to the final pit selections, considered the mining of larger pits at higher gold prices and associated risks. This is mainly driven by ounces, changes in strip ratio, life of the pit, and the value of the pits at different metal prices. As an example, Table 15-26 shows the revenue for the $1,000/oz to $1,200/oz gold price pits and the potential net cash flow generated for each price scenario for Sessenge.

18^th September 2018 Page 255

Kibali Gold Mine NI 43-101 Technical Report

Table 15-26 Sessenge Gold Price versus Pit Size and Sell Price Risk Analysis

Sell Price ($/oz)	Operating Cost ($ ‘000)	Revenue ($ ‘000)	Cash Flow ($ ‘000)
$1,100/oz Pit Sensitivities on Sell Price
1000	-166,751	271,579	104,828
1050	-166,751	285,158	118,407
1100	-166,751	298,737	131,986
1150	-166,751	312,316	145,565
1200	-166,751	325,895	159,144
1250	-166,751	339,474	172,723
1300	-166,751	353,053	186,302
1350	-166,751	366,632	199,881
1400	-166,751	380,211	213,460
1450	-166,751	393,789	227,039
1500	-166,751	407,368	240,618
$1,000/oz Open Pit Sensitivities on Sell Price
1000	-153,316	258,901	105,585
1050	-153,316	271,846	118,530
1100	-153,316	284,791	131,475
1150	-153,316	297,736	144,420
1200	-153,316	310,681	157,365
1250	-153,316	323,626	170,310
1300	-153,316	336,571	183,255
1350	-153,316	349,517	196,200
1400	-153,316	362,462	209,146
1450	-153,316	375,407	222,091
1500	-153,316	388,352	235,036
$1,200/ oz Open Pit Sensitivities on Sell Price
1000	-178,772	281,686	102,913
1050	-178,772	295,770	116,997
1100	-178,772	309,854	131,082
1150	-178,772	323,938	145,166
1200	-178,772	338,023	159,250
1250	-178,772	352,107	173,335
1300	-178,772	366,191	187,419
1350	-178,772	380,276	201,503
1400	-178,772	394,360	215,587
1450	-178,772	408,444	229,672
1500	-178,772	422,528	243,756

Figure 15-15 illustrates the value/tonnage curves that were generated for the Sessenge pit to illustrate the net cash flow at different gold prices.

18^th September 2018 Page 256

Kibali Gold Mine NI 43-101 Technical Report

Figure 15-15 Sessenge Value/Tonnage Curve

The value / tonnage curve for Sessenge is a relatively flat curve between the $1,000/oz pit and the $1,200/oz pit.

Choosing the $1,000/oz pit and selling at a price of $1,300/oz gives a cash flow of $183 M whereas the $1,200/oz pit, at the same selling price of $1,300/oz, will return $187 M. The value difference of $4 M is thus the opportunity gain of mining a bigger pit should the gold price remain at $1,300/oz for the entire duration of the Sessenge pit.

Conversely, if the gold price drops to $1,000/oz then the ounces of gold produced from the larger $1,200/oz pit, sold at $1,000/oz, generates a value of $103 M while the $1,000/oz pit, at the same price, generates a value of $106 M.

The risks and opportunities, in value terms, are the same so the smaller pit was chosen for Sessenge. This approach was applied to the selection of all Ore Reserve pits.

18^th September 2018 Page 257

Kibali Gold Mine NI 43-101 Technical Report

15.7 Mine Design

Open Pits

The selected pit shells were used as guidelines to design the practical ultimate pits with internal phases. Pit design parameters were selected based on the overall pit geometry, geotechnical data and information, and the mine production rate. Pit and internal phases were designed using Surpac software, integrating the recommended standards for road width and minimum mining width based on an efficient operations for the size of mining equipment chosen for the open pit operations.

Comparisons of 2017 Whittle shells to 2016 shells and 2016 reserve pits were completed to assess the changes. The KCD, Sessenge, Kombokolo, and Pamao pits were redesigned as the changes were deemed relevant.

All designs were based on approved geotechnical slope angles provided by the Geotechnical department and consultants; these have been detailed previously in this Section under Slope Angles, and they are summarised again in Table 15-27.

Table 15-27 Summary of Pit Design Parameters

Material	Bench Height (m)	Berm Width (m)	Batter Angle (°)	Inter Ramp Slope Angle (°)
Weathered	5-10	4-5	27-50	27-50
Transition	10	4-6	27-65	38-55
Fresh Rock	10-20	4-6	50-80	48-59

The two-staged Pakaka pit design strategy was maintained for 2017 (Figure 15-16). Mining in Pakaka for 2017 was concentrated on pushback 1 and is expected to be mined out in Q1 2018. The final pushback has been planned for 2020.

All designs were checked to ensure that they provided enough width for the mining fleet in order to avoid constraints and difficulties during excavations. The selected and designed pushbacks ensured adequate waste deferral in the early stages and provided a continuous supply of the appropriate blend of ore to the plant.

The year end 2017 Kibali open pit designs are shown in Figure 15-16 to Figure 15-22.

18^th September 2018 Page 258

Kibali Gold Mine NI 43-101 Technical Report

Figure 15-16 Pakaka Pushback Designs

Figure 15-17 KCD End of 2017 Reserve Pit

18^th September 2018 Page 259

Kibali Gold Mine NI 43-101 Technical Report

Figure 15-18 Sessenge E End of 2017 Reserve Pit

Figure 15-19 Pakaka End of 2017 Reserve Pit

18^th September 2018 Page 260

Kibali Gold Mine NI 43-101 Technical Report

Figure 15-20 Kombokolo End of 2017 Reserve Pit

Figure 15-21 Pamao End of 2017 Reserve Pit

18^th September 2018 Page 261

Kibali Gold Mine NI 43-101 Technical Report

Figure 15-22 Gorumbwa End of 2017 Reserve Pit

Underground Mine

A significant portion of the capital and access development for the underground mine is in place. To date, 27 km of capital development and 10 km of waste access development has been completed. The current LOM predicts a further 21 km of lateral capital development and 18 km of waste access development.

The items of key capital infrastructure remaining to be developed are the 9101 decline, 9101 incline, southern exhaust raises and the 3101 / 3102 access development. This capital infrastructure requires additional resource infill and grade control diamond drilling before finalising the design. Figure 15-23 shows the current (December 2017) mine as-built (grey) and the LOM development.

18^th September 2018 Page 262

Kibali Gold Mine NI 43-101 Technical Report

Figure 15-23 Life of Mine Development and As Built (Dec 2017)

The key components of the materials handling system (Figure 15-24) are:

☐ Teleremote (and manually operated) loaders tramming from the stopes and development faces to ore pass finger raises on that level.

☐ Eight raise bored ore passes with finger raises on production levels.

☐ Haulage level (210 level) – up to three remote operated automated loaders tram ore from the passes to two grizzlies.

☐ Two coarse ore bins.

☐ Two crushers.

☐ Two fine ore bins.

☐ Conveyor transport of ore from crushed ore bin to shaft loading pocket.

☐ Shaft haulage (740 m deep).

☐ Headframe ore bin.

☐ Conveyor haulage from shaft to process plant (including facility to place waste on an interim stockpile).

18^th September 2018 Page 263

Kibali Gold Mine NI 43-101 Technical Report

Figure 15-24 Materials Handling System

The materials handling system is also supplemented by truck haulage up the decline to the ROM pad at the process plant.

15.8 External Audits

The December 2013 Kibali underground and open pit Ore Reserves were independently reviewed by Snowden Mining Industry Consultants (Snowden). Snowden found no critical action items but did recommend continuous improvement actions relating to the underground Ore Reserves and these have been addressed by Kibali Goldmines.

The 31st December 2017 Kibali Mineral Resources and Ore Reserves estimates are currently being reviewed by Optiro Pty Ltd. Optiro has provided the following overview of their audit:

The Ore Reserves have been assessed for JORC and SAMREC Code compliance. Optiro considers that the processes supporting the Ore Reserves, as reviewed in this document, are at a level commensurate with accepted through to best industry practice.

18^th September 2018 Page 264

Kibali Gold Mine NI 43-101 Technical Report

16 Mining Methods

The Kibali Mine comprises both open pit and underground mining operations. The general layout of the mine is shown in Figure 16-1.

Figure 16-1 Plan Showing Relative Positions of Open Pits and Main Mine Infrastructure

16.1 Open Pit Method

Open pit mining is carried out using conventional drill, blast, load and haul surface mining methods. Mining of the main pits is carried out by a mining contractor, KMS; in addition, some of the satellite pits have been mined by local Congolese contractors. In 2017, the total production of ore and waste from pits was 34.0 Mt, with an average stripping ratio of 5.9. Approximately 5.0 Mt of ore at an average grade of 2.36 g/t Au containing a total of 376 koz of in situ gold was mined in 2017.

Historical production from the Kibali open pits, up to 2017, is detailed in Table 16-1.

18^th September 2018 Page 265

Kibali Gold Mine NI 43-101 Technical Report

Table 16-1 Kibali Open Pits Historical Production

Open Pit	Unit	2012	2013	2014	2015	2016	2017
KCD	S/R	63.0	4.7	4.3	4.2	1.1	3.3
	Ore (kt)	86	4,335	5,516	4,458	764	366
	Waste (kt)	5,437	20,199	23,782	18,849	859	1,207
Sessenge	S/R	-	-	-	-	-	-
	Ore (kt)	-	-	-	-	-	-
	Waste (kt)	-	-	-	-	-	-
Pakaka	S/R	-	-	-	-	4.8	4.2
	Ore (kt)	-	-	-	-	2,350	3,386
	Waste (kt)	-	-	-	-	11,173	14,080
Mengu Hill	S/R	-	-	-	3.2	8.1	1.9
	Ore (kt)	-	-	-	1,511	1,180	441
	Waste (kt)	-	-	-	4,888	9,541	855
Komnokolo	S/R	-	-	-	-	12.8	18.5
	Ore (kt)	-	-	-	-	278	686
	Waste (kt)	-	-	-	-	3,548	12,722
Pamao	S/R	-	-	-	-	-	-
	Ore (kt)	-	-	-	-	-	-
	Waste (kt)	-	-	-	-	-	-
Gorumbwa	S/R	-	-	-	-	-	-
	Ore (kt)	-	-	-	-	-	-
	Waste (kt)	-	-	-	-	-	-
Mofu/Rhino	S/R	-	-	7.3	4.9	3.4	2.6
	Ore (kt)	-	-	95	89	76	85
	Waste (kt)	-	-	693	440	258	225
Total	S/R	63.0	4.7	4.4	4.0	5.5	5.9
	Ore (kt)	86	4,335	5,611	6,058	4,647	4,964
	Waste (kt)	5,437	20,199	24,475	24,176	25,379	29,089
	Total (kt)	5,524	24,534	30,086	30,235	30,026	34,053
	Mined Grade (g/t Au)	1.75	2.52	3.15	3.89	2.86	2.36

From 2018 onwards, open pit production will come from the KCD, Sessenge, Pakaka, Kombokolo, Pamao, and, Gorumbwa deposits. The Mengu Hill, Mofu, and Rhino pits were previously depleted, as of 2017. The estimated LOM production of ore and waste for the open pits schedule are as presented in Table 16-2.

Table 16-2 Kibali Open Pits, Reserves Basis

Open Pit	Ore		Waste	Total
	Tonnes (kt)	Grade (Au g/t)	Tonnes (kt)	Tonnes (kt)	Strip Ratio
KCD	5,426	2.12	21,749	27,176	4
Sessenge	3,949	2.39	12,685	16,634	3.2
Pakaka	3,788	2.59	31,033	34,821	8.2
Kombokolo	1,395	3.16	10,124	11,519	7.3
Pamao	3,434	1.76	7,951	11,384	2.3
Gorumbwa	4,150	2.81	40,644	44,795	9.8
Total	22,142	2.38	124,187	146,329	5.6

Figure 16-1 shows a longitudinal section of Pushback 3 in the KCD pit. Figure 16-2 illustrates a longitudinal section of Pushback 2 in the Pakaka pit and Figure 16-3 presents a longitudinal section of Pushback 1 and 2 in the Gorumbwa pit.

18^th September 2018 Page 266

Kibali Gold Mine NI 43-101 Technical Report

Figure 16-2 Long Section of the KCD Pit and Pushback 3

Figure 16-3 Long Section of the Pakaka Pit and Pushback 2

18^th September 2018 Page 267

Kibali Gold Mine NI 43-101 Technical Report

Figure 16-4 Long Section of the Gorumbwa Pit and Pushback 2

Open Pit Mining Equipment

During 2017, local mining contractors were used to assist the main contractor, KMS, to strip waste in the Kombokolo and Pakaka pits and also for the completion of the Rhino pit. Commencement of the KCD pushback 3 pit and the undulating topography in the pit required the use of smaller equipment until wider benches were generated to enable the use of larger CAT 777 trucks and larger shovels that are used by KMS in its normal mining operations.

The mining fleet is presented in Table 16-3. The fleet size for 2018 and beyond is projected to remain fairly consistent with no further material additions of equipment being required. The maintenance schedule allows for some annual rebuilds of the equipment each year. The fleet size is considered to be adequate to achieve the LOM production targets.

Table 16-3 Current Primary Open Pit Mine Equipment Fleet

Fleet	Current Quantity	Expected
	2017	2018
Liebherr 9350 Excavators	3	3
Liebherr 9350 Excavators	2	2
CAT 777G Dump Trucks	23	23
CAT 992 Wheel Loaders	2	2
CAT D9R Dozers	8	8
CAT 16M Graders	3	3
CAT 834M Pushers	2	2
Blast Drill Rigs	8	8
Water Bowsers	2	2

No local Congolese mining contractors are envisaged for 2018 and beyond.

18^th September 2018 Page 268

Kibali Gold Mine NI 43-101 Technical Report

Waste Dumps

An estimated 124 Mt of waste will be mined over the remaining LOM.

The capacity of the Kibali open pit waste dumps has been evaluated based on the latest pit designs to confirm that there is adequate dump capacity for the estimated LOM tonnage of waste. A swell factor of 30% was considered in all waste dump capacity evaluations. Haul roads were also adjusted, where necessary, to ensure they provide easy access where pit ramps are day lighting.

No in pit dumping was carried out in 2017 and none is planned for 2018, as the KCD and Sessenge mines continue to explore the potential deposit of the KCD 3000 lode up plunge. Future work will, however, consider the use of some of the satellite pits for waste disposal, based on the mining sequence.

Table 16-4 summarises the related waste dump capacities.

Table 16-4 Waste Dump Capacities

Waste Dump	Design Capacity (Mm3)	Waste Dumped to Date (Mm3)	Planned Future Waste (Mm3)
KCD/Sessenge	45.93	23.05	14.84
Pakaka	28.11	11.37	12.38
Kombokolo (Dumps 1 and 2)	13.97	8.24	3.76
Pamao (Dumps 1 and 2)	15.21	0	5.37
Gorumbwa	19.19	0	14.4
Total	122.41	42.66	50.75

16.2 Underground Method

The KCD underground mine is designed around the use of long hole stoping with a planned production rate of 3.6 Mtpa.

Development of the underground mine commenced in 2013. Stoping commenced in 2015, with the production ramping up to approximately 1.8 Mtpa by 2017. Full production of 3.6 Mtpa is forecast to be achieved in 2018.

Initial access and mine production have been via a twin decline system from surface. In 2017 the production shaft (740 m deep) and materials handling system were commissioned, and from 2018, the majority of ore will be hoisted up the shaft. In future the surface decline system will be used to haul from some of the shallower zones and to supplement the shaft hoisting capacity. Table 16-5 details historical KCD underground production to the end of 2017.

18^th September 2018 Page 269

Kibali Gold Mine NI 43-101 Technical Report

Table 16-5 Kibali KCD Underground Historical Production

Unit	2013	2014	2015	2016	2017
Trucked Ore (kt)	-	96	812	1,586	1,669
Shaft Ore (kt)	-	-	-	-	119
Total Ore (kt)	-	96	812	1,586	1,788
Mined Grade (g/t Au)	-	3.68	5.03	4.71	5.51
Development Waste (kt)	125	607	739	58	706

Four main mineralised zones contribute to the bulk of the Ore Reserve, these are the 5101, 5102, 9101, and 9105. Five other mineralised zones, 3101, 3102, 5104, 5105, and 5110 - contribute the remaining 12% of the Ore Reserve (Figure 16-5).

Figure 16-5 KCD Underground Mining Zones

The proposed mining methods are variants of long hole open stoping with cemented paste:

☐ Primary / Secondary long hole open stoping (primary 28% of Ore Reserve tonnes, secondary 40% of tonnes).

☐ Advancing face long hole open stoping (23% of tonnes).

☐ Longitudinal open stoping (9% of tonnes).

Mining methods are reviewed periodically as further resource infill and grade control drilling changes the shape of the ore zones, however, there have been no significant changes to the mining methods.

A long section (looking NW) showing the currently anticipated mining methods is shown in Figure 16-6.

18^th September 2018 Page 270

Kibali Gold Mine NI 43-101 Technical Report

Figure 16-6 Long Hole Open Stoping Methods

Primary / secondary stoping is used in the wider zones. The level interval is 35 m (floor to floor) and stopes are mined either as single or multiple lifts (up to four lifts), depending upon stope geometry and stable span analysis. Primary stopes are typically 20 m along strike and secondary stopes are typically 30 m along strike. The width of primary stopes can be up to approximately 40 m across strike. The controlling span for primary stope size is generally the side (north and south) rock walls. Secondary stopes are up to 30 m across strike. The controlling span for secondary stope size is generally the side wall paste exposure of the adjacent primary stopes.

Where the deposit is too wide for a single stope span (>30 m to 40 m wide), multiple primary and secondary stopes are mined retreating from hangingwall to footwall. The first stope is paste-filled prior to mining of the adjacent stopes. A slot raise is developed by raise boring. Production drillholes are down holes of 76 mm or 90 mm diameter.

Advancing face transverse stoping is used in the 9101 zone, which has a shallow plunge (20° to 30°) to the NE. The level interval varies from 25 m to 35 m to optimise extraction. Typically, the stopes are 25 m down plunge and 25 m across plunge. Stopes are mined as a single lift or multiple lifts (up to three lifts per stope), depending upon the ore zone thickness. Stopes are paste filled prior to the mining of adjacent stopes. A slot raise is developed by raise boring. Production drillholes are either up or down holes of 76 mm or 90 mm diameter.

Longitudinal stoping is used in the narrow ore zones (below 15 m width). In the steeper areas (over 60°) the level interval varies from 20 m to 35 m. In the flatter areas (5° to 60° dips) ore drives are located on the footwall and the level interval is controlled by the dip, minimising footwall waste, and limiting stope width (up dip) to 20 m. Stopes are paste-filled prior to mining of adjacent stopes to maintain hangingwall stability.

18^th September 2018 Page 271

Kibali Gold Mine NI 43-101 Technical Report

Underground Mining Equipment

The underground production and development operations are presently being undertaken by a contractor (Byrnecut). The underground equipment is owned by Kibali and currently maintained and operated by Byrnecut in a purpose build workshop on the surface. All maintenance is carried out on surface, as there are no underground workshops.

Byrnecut will continue as the contract mining company until July 2018 when Kibali Goldmines will assume full responsibility for the operations, production and development, maintenance, and dewatering, etc.

The equipment fleet over the mine life is summarised in Table 16-6.

Table 16-6 Kibali Underground Mining Equipment

Unit	Current     (2018)	2023	2028	2032   (End of LOM)
Sandvik DL421C Production Rig	4	4	3	1
Sandvik DD421C Development Rig (Jumbo)	4	2	1	1
Sandvik LH621 Production Loader	13	13	13	10
Sandvik TH551Truck (51t)	8	6	5
Charge Machines	4	5	5	1
ITs	6	6	5	2
Grader	2	2	2	2
Agi Truck	1	1	1	1
Cable Bolter	2	2	2	1
Rock Breaker	1	1	1	1
Total Fleet	45	42	38	20

All underground equipment is equipped with ANSUL or OEM fire suppression systems and handheld fire extinguishers.

16.3 Life-of-Mine Plan

Open Pits

Production Scheduling

The six Ore Reserve pit designs were scheduled with their respective updated block models with signed off cut-off grades in MineSched software for the 2018 Life of Mine schedule and budget. The mine schedule was based on a marginal grade cut-off. Material classes for various material types were created with different grade categorisation of high, medium and low-grades. These categorisations were based on the grade and tonnage distribution of ore in each deposit.

The mine schedule was generated based on historic rainfall patterns and scheduled calendar days. With nine months of rainfall expected in the year, the monthly budget was aligned to this to cater for lost days as a result of heavy rainfall. New pits are usually brought into production between December and March so as to be able to mine the saprolite in dry periods. Sheeting (with fresh

18^th September 2018 Page 272

Kibali Gold Mine NI 43-101 Technical Report

waste rocks) of haul roads, ramps, and pit floors is practiced where possible to keep haul trucks running in wet conditions.

Figure 16-7 shows analysis of rainfall pattern and lost hours from rain over a four-year period. Mining operations are carried out seven days per week, three shifts per day, utilising four shift crews.

Figure 16-7 Historic Rainfall Pattern and Lost Production Hours from Rain

A part of the KCD pit schedule includes 61 kt of Inferred Mineral Resources for a total of 3.3 koz of gold that has been included in the LOM plan, but excluded from the Ore Reserve estimate. No other Inferred mineralisation has been included in the LOM. As shown in Table 16-7, this comprises 1% of the entire open pit scheduled ore and poses no threat to the business plan.

Table 16-7 Mine Plan Material Classification Risk

Item	Pakaka	Kombokolo	KCD-PB3	Sessenge	Total	% Material
Proved
Ore Tonnes (kt)	1070	526	171	-	803	21%
Ore Grade (g/t)	4.52	3.45	2.58	-	3.41
Proved (koz	15	58	14	-	88	28%
Probable
Ore Tonnes (kt)	-	869	1,450	619	2,938	77%
Ore Grade (g/t)	-	2.98	2.17	2.11	2.40
Probable (koz)	-	83	101	42	226	71%
Inferred
Inferred Tonnes (kt)	-	-	61	-	61	2%
Inferred Grade (g/t)	-	-	1.70	-	1.70
Inferred (koz)	-	-	3	-	3	1%
Total Open Pit
Total Ore Tonnes (kt)	1070	1,395	1,682	619	3,803	100%
Grade (g/t)	4.52	3.16	2.19	2.11	2.60
Total (koz)	15	142	119	42	318	100%
Pit %
	3%	37%	44%	16%	100%

18^th September 2018 Page 273

Kibali Gold Mine NI 43-101 Technical Report

Any Inferred material identified in any of the pits has been flagged to the Mineral Resources Department for grade control drilling to allow potential conversion into Indicated Mineral Resources and subsequently into Ore Reserves, creating opportunities to improve the business plan.

Pit Sequencing

Changes were made to the Kibali open pit mine sequencing as a consequence of the drilling during the year. The KCD PB3 pit was brought forward to August 2017 after confirming metallurgical recoveries and accurately defining voids from old artisanal workings. The introduction of this pit pushed out the commencement of the Gorumbwa pit, which is now scheduled between 2019 and 2023 and subsequently the Pamao pit is now planned between 2021 and 2024. The Pakaka final pushback has also been subsequently moved out to 2022.

Results from the Sessenge grade control drilling better defined the higher-grade shoots of the deposit and accordingly this pit has also been rescheduled to be mined at the end of Q1 2018.

Open pit production rates for the schedule were lowered in 2018 and onwards, based on the expectations of higher-grade underground production from the commissioning of the underground shaft. 24.6 Mt total ore and waste tonnage is expected to be mined in year 2018. This will ensure adequate cash flow and avoid locking working capital in excessive stockpiles. Opportunities exist with the Inferred material within the current pits that may be converted to Ore Reserves with additional drilling.

A ramp-down of open pit production begins from 2023 with an extraction rate of 11.6 Mtpa. The open pit end of life is estimated for 2026 based on current Ore Reserves, as shown in Figure 16-8 and Table 16-8.

Figure 16-8 Kibali Open Pit Mining Rate

18^th September 2018 Page 274

Kibali Gold Mine NI 43-101 Technical Report

Table 16-8 Open Pits Mining Sequence Over the LOM

Open Pit	Unit	2018	2019	2020	2021	2022	2023	2024	2025	2026	Total
	S/R	4.5	3.7	4.4	3.3	1.4	-	-	-	-	4.0
KCD	Ore (kt)	1,682	1,815	1,271	579	78	-	-	-	-	5,426
	Waste (kt)	7,498	6,626	5,597	1,922	106	-	-	-	-	21,749
	S/R	4.0	3.4	3.8	2.4	1.7	-	-	-	-	3.2
Sessenge	Ore (kt)	619	1,071	1,060	743	457	-	-	-	-	3,949
	Waste (kt)	2,471	3,610	4,035	1,808	761	-	-	-	-	12,685
	S/R	6.6	-	-	-	375.4	21.6	5.8	3.7	1.9	8.2
Pakaka	Ore (kt)	107	-	-	-	15	198	671	1,474	1,323	3,788
	Waste (kt)	698	-	3,862	4,528	5,706	4,288	3,884	5,489	2,578	31,033
	S/R	7.3	-	-	-	-	-	-	-	-	7.3
Kombokolo	Ore (kt)	1,395	-	-	-	-	-	-	-	-	1,395
	Waste (kt)	10,124	-	-	-	-	-	-	-	-	10,124
	S/R	-	-	-	3.2	1.4	3.6	2.0	-	-	2.3
Pamao	Ore (kt)	-	-	-	344	933	754	1,402	-	-	3,434
	Waste (kt)	-	-	-	1,089	1,322	2,707	2,832	-	-	7,951
	S/R	-	8.4	6.7	17.4	14.2	4.2	-	-	-	9.8
Gorumbwa	Ore (kt)	-	1,178	859	599	818	695	-	-	-	4,150
	Waste (kt)	-	9,886	5,789	10,418	11,598	2,953	-	-	-	40,644
TOTAL	S/R	5.5	5.0	6.0	8.7	8.5	6.0	3.2	3.7	1.9	5.6
	Ore (kt)	3,803	4,064	3,190	2,266	2,301	1,647	2,074	1,474	1,323	22,142
	Waste (kt)	20,792	20,122	19,283	19,765	19,494	9,947	6,717	5,489	2,578	124,187
	Total (kt)	24,595	24,186	22,473	22,031	21,795	11,594	8,790	6,963	3,901	146,329
	Mined Grade (g/t Au)	2.6	2.31	2.21	2.62	2.18	2.49	2.24	2.74	2	2.38
	Total In-Situ (koz)	318	302	226	191	161	132	149	130	85	1,694

18^th September 2018 Page 275

Kibali Gold Mine NI 43-101 Technical Report

Underground Mine

Production Schedule

The LOM schedule for the KCD underground mine has been created in Datamine 5DP / EPS software. It is a task-based dependency schedule. The overall Kibali LOM has been created in Excel by combining the output from the underground and open pit scheduling packages. The underground physicals by year are shown in Table 16-9.

The LOM schedule is illustrated in five steps at two-year intervals, followed by the end of LOM in Figure 16-9 to Figure 16-14 inclusive. This LOM plan was prepared prior to the completion of the 31st December 2017 Ore Reserve and there are some differences between the designs and physical quantities in the 2017 Ore Reserve estimate.

The KCD underground mine sustains a production rate of 3.6 Mtpa for 10 years. This LOM has a long tail of declining production over a further nine years. The schedule is expected to be progressively optimised to extend the period of 3.6 Mtpa production rate.

18^th September 2018 Page 276

Kibali Gold Mine NI 43-101 Technical Report

Table 16-9 Kibali KCD Underground Life of Mine Physicals

	2018	2019	2020	2021	2022	2023	2024	2025	2026	2027	2028	2029	2030	2031	2032	2033	2034	2035	2036	Total
TOTAL ORE
Total Ore (kt)	3,661	3,642	3,620	3,601	3,599	3,680	3,692	3,636	3,664	3,491	2,533	1,236	595	596	545	845	506	503	207	44,134
Total Grade (g/t)	4.82	4.88	5.13	4.71	5.52	5.3	5.15	4.86	4.85	5.14	5.16	4.31	7.4	6.38	4.97	4.41	4.18	4.79	4.82	5.04
Total Ounces (koz)	567	571	597	546	639	627	611	568	572	577	420	171	142	122	87	120	68	77	32	7,156
STOPE ORE
Stope Ore (kt)	3,413	3,494	3,271	3,300	3,380	3,669	3,680	3,626	3,615	3,491	2,533	1,236	595	596	545	845	506	503	207	42,763
Stope Grade (g/t)	4.79	4.84	5.13	4.68	5.54	5.31	5.15	4.86	4.85	5.14	5.16	4.31	7.4	6.38	4.97	4.41	4.18	4.79	4.82	5.04
Stope Ounces (koz)	526	543	539	497	602	626	609	567	564	577	420	171	142	122	87	120	68	77	32	6,926
DEVELOPMENT ORE
Development Ore (kt)	248	148	349	301	219	10	12	11	50											1,371
Development Grade (g/t)	5.18	5.9	5.2	5	5.19	4.13	5.19	4.79	5.07											5.22
Development Ounces (koz)	41	28	58	48	37	1	2	2	8											230
DEVELOPMENT WASTE
Total Development Waste (kt)	684	658	658	667	143	10	2	5	49											2,998
DEVELOPMENT LATERAL
Capital (m)	5,409	5,926	4,160	4,131	418				85											20,684
Normal Ore (m)	3,061	1,863	4,487	3,782	2,841	133	156	140	657											17,411
Normal Waste (m)	3,476	2,687	4,709	5,022	1,537	137	20	62	583											18,747
Production (m)	1,646	1,021	720	1,094	1,111	1,029	1,256	820	897	863	849	525	255	120	195	165	165	105	30	13,825
Total Lateral (m)	13,592	11,498	14,075	14,029	5,907	1,299	1,432	1,022	2,222	863	849	525	255	120	195	165	165	105	30	70,667
DEVELOPMENT VERTICAL
Capital (m)	1,503	1,229	363	132	37															5,997
Normal (m)
Production (m)	2,367	1,896	1,460	1,543	1,888	1,644	1,696	1,299	1,145	1,085	1,040	850	392	273	390	301	359	210	60	21,360
Total Vertical (m)	3,870	3,125	1,823	1,675	1,925	1,644	1,696	1,299	1,145	1,085	1,040	850	392	273	390	301	359	210	60	27,357
DESTINATION SHAFT
Shaft Ore (kt)	2,195	2,318	3,162	3,212	3,281	3,456	3,527	3,585	3,664	3,491	2,533	1,236	595	596	545	845	506	503	207	39,586
Shaft Grade (g/t)	5.15	5.1	5.25	4.8	5.54	5.37	5.18	4.88	4.85	5.14	5.16	4.31	7.4	6.38	4.97	4.41	4.18	4.79	4.82	5.12
Shaft Ounces (koz)	363	380	533	496	584	597	587	562	572	577	420	171	142	122	87	120	68	77	32	6,511
Total Shaft Tonnes (kt)	2,402	2,493	3,414	3,509	3,436	3,553	3,622	3,669	3,861	3,575	2,606	1,282	624	122	559	858	519	511	209	41,514
DESTINATION PORTAL
Portal Ore (kt)	1,466	1,323	458	389	319	224	166	51												4,549
Portal Grade (g/t)	4.32	4.49	4.34	3.98	5.37	4.27	4.49	3.88	3.54											4.42
Portal Ounces (koz)	204	191	64	50	55	31	24	6												646
Total Portal Tonnes (kt)	2,064	1,914	926	877	391	234	168	51	8											6,908
PRODUCTION
Total Production Drill (km)	380	256	208	205	266	254	262	232	183	151	116	76	32	43	39	22	41	28	10	2,982
Total Backfill Paste (kt)	2,327	2,523	2,378	2,361	2,368	2,853	2,501	2,528	2,708	2,630	1,942	1,079	545	350	403	573	369	304	303	31,083
Total Cement Tonnes (kt)	109	111	113	87	94	96	95	89	97	86	62	39	20	12	14	20	13	11	11	1,180

18^th September 2018 Page 277

Kibali Gold Mine NI 43-101 Technical Report

Figure 16-9 KCD Underground LOM on 1st Jan 2019

Figure 16-10 KCD Underground LOM on 1st Jan 2021

18^th September 2018 Page 278

Kibali Gold Mine NI 43-101 Technical Report

Figure 16-11 KCD Underground LOM on 1st Jan 2023

Figure 16-12 KCD Underground LOM on 1st Jan 2025

18^th September 2018 Page 279

Kibali Gold Mine NI 43-101 Technical Report

Figure 16-13 KCD Underground LOM on 1st Jan 2027

Figure 16-14 KCD Underground LOM on 1st Jan 2035

18^th September 2018 Page 280

Kibali Gold Mine NI 43-101 Technical Report

16.4 Underground Mine Infrastructure and Services

This Section is specific to the infrastructure directly associated with underground mine operations. For all other general infrastructure located at surface, see Section 18 (Project Infrastructure).

Ore and waste are hoisted to surface via a concrete-lined vertical shaft, completed during 2015. The shaft has an internal diameter of 8 m and is equipped with two 14.5 t skips and a service cage for man access. Design capacity of the shaft is 3.6 Mtpa, in excess of the planned average annual production of 3.3 Mtpa.

The development and installation of the underground mine infrastructure (ventilation fans, pumping, electrical systems, water supply, and fire protection) has been completed. The haulage level will be completed in 2019.

The material handling system comprises eight ore passes feeding to the haulage drive at the 5300 mRL level. Ore is fed into two primary jaw crushers, each provided with 1,000 t capacity coarse and fine ore bins above and below them. Conveyors transfer the ore to the skip loading boxes at the shaft. The skip loading system is fully automated to transfer and load ore into the skips.

The current ventilation system at Kibali consists of three exhaust raises, each equipped with a Zitron 800 kW axial flow exhaust fan. Total air extracted through the three raises is 800 m3/s. Fresh air enters the mine through the declines, two intake raises and the production shaft. An additional exhaust raise is planned for 2019, which will be equipped with further 800 kW Zitron fan, increasing the air volume to 1,200 m³/s.

Other underground infrastructure includes:

☐ Small workshop for daily equipment servicing.

☐ Explosives magazine.

☐ Offices and small mechanical/electrical workshops.

☐ Electrical ring feed and transformers. Both feed cables are located in the hoist shaft.

☐ Raw water mains to supply water at 20 l/s.

☐ Sump with settling chamber.

☐ Principal pump station with two parallel pumping systems for clean water including two rising mains to the surface. Both rising mains are located in the haulage shaft.

☐ Two vertical plate filters at each crusher to process the mud from the settling chamber. The filter cake is conveyed to the fine ore bins.

☐ Paste backfill distribution system from the surface backfill plant, with a design placement capacity of 200 m3/hr. Delivery lines from surface are installed in dedicated raises.

The 11 kV power supply for underground is distributed in a ring main system that provides two sources of power supply at all times. Back-up generators have been installed on the surface to provide power to the main pumps and primary ventilation fans in case of a failure in the power supply to the shaft.

18^th September 2018 Page 281

Kibali Gold Mine NI 43-101 Technical Report

There are three main pump stations pumping water out of the mine:

☐ C615 Main pump station: six 110 kW Challenge pumps; maximum pump rate 120 l/s at 245 m head.

☐ B1 decline 580 pump station: two 90 kW Flygt pumps; maximum pump rate 75 l/s.

☐ Shaft temporary station at production level: two 110 kW Scamont pumps; maximum pump rate 16 l/s.

The main shaft pump station at the crusher level is equipped with three 1,140 kW Scamont pumps, each with a 120 l/s capacity at a 720 m head. Only two pumps will be operated at any one time, with the third kept on standby. The shaft pump capacity is 240 l/s via two separate steel rising mains to the surface for the pumps.

The underground mine is equipped with a fibre optic communication system throughout the mine and a high-speed Ethernet system in the vertical shaft. All fixed equipment is linked to this system and fed into a central control room. This facilitates continuous monitoring, control, and alarms.

A seismic monitoring system is being installed in three phases as the mine develops. The first phase was fully commissioned in February 2018. The planned sensor array consists of 12 triaxial 4.5 Hz geophones which will provide coverage of the entire KCD deposit. The first phase consists of six sensors providing a baseline of seismic activity before full stope production is ramped-up.

In the opinion of the QP, the infrastructure is adequate and has been, or is being, provided at Kibali to support the anticipated production targets from the underground mine.

18^th September 2018 Page 282

Kibali Gold Mine NI 43-101 Technical Report

17 Recovery Methods

17.1 Processing Plant

Kibali ore is blended using both KCD underground ore plus ore sourced from satellite open pits at Kibali. The process plant has been treating Kibali KCD underground ore since 2015 and has demonstrated reasonably consistent recovery performance. The flow sheet comprises crushing, ball milling, classification, gravity recovery, a conventional CIL circuit, flash flotation, also conventional flotation, together producing a concentrate which goes to ultra-fine-grinding, and a dedicated intensive cyanide leach. This process consists of industry standard technology and is appropriate for Kibali’s style of mineralisation. The Kibali gold processing plant comprises two largely independent processing circuits, the first one designed for oxide, transition and free milling ore sources and the second for sulphide refractory ore. However, both circuits are designed to be switched to process sulphide ore when the oxide, transition and free milling ore sources have been depleted. A simplified flowsheet can be seen in Figure 17-1.

The oxide ore is recovered through a standard crushing, milling, and gravity plus CIL operation.

The sulphide ore requires: crushing; milling; flotation; ultra-fine grinding (UFG); a pumpcell circuit preceded by a three-tank gravity flow pre-oxidation circuit to passivate cyanide consuming sulphides as well as liberate the gold. The first two tanks are subject to highly intensive oxidation with cyanide being introduced into the third to fifth tanks for pre-leaching, where the resultant product gravitates to a pumpcell Carbon-in-Pulp (CIP) circuit with high concentrations of activated carbon. The pumpcell residue stream may still contain some residual gold which is then pumped to the main CIL circuit for final leaching to scavenge the remaining leachable gold. The flexibility of the plant design allows for an extended pre-oxidation and pre-leach step within the CIL occurring after the initial pre-oxidation circuit but prior to the stream being routed to the pumpcell circuit.

Most of the ore bodies contain some extent of free native gold, which means it is large enough to recover via a density separating step which is performed with Knelson gravity concentrators during the milling cycle.

The processing plant rated throughput is 3.6 Mtpa of soft oxide rock ore through the oxide circuit and 3.6 Mtpa of primary sulphide rock ore through a parallel sulphide circuit. Once the plant is sulphide only, the capacity is 7.2 Mtpa of sulphide ore. Kibali’s operational performance has demonstrated that the process plant is fully capable of its design capacity, and further modifications to the mills with an increased motor size coupled with a decreased inlet trunnion size has allowed for an even greater power draw and hence higher throughputs. The 2018 process feed plan can be found in Table 17-1.

18^th September 2018 Page 283

18th September 2018 Page 284

Kibali Gold Mine NI 43-101 Technical Report

Table 17-1 2018 Process Feed Plan

Type	Tonnes Ore     (kt)	Grade (g/t     Au)	Contained     Metal (koz)
Sulphide	6,128	3.89	766
Transition + Soft Sulphide	798	2.43	62
Oxide	708	2.45	56
Total	7,627	3.6	884
Recovery (%)		87.52
Gold Production (koz)		773

The plant has the capacity to make the stated through-put based on historic throughput for the oxides and sulphides. The graph depicted in Figure 17-2 attests to the continued improvement in throughput ultimately well beyond design capacity. No fatal flaws have been determined.

Figure 17-2 Kibali Plant Performance – Tonnes Treated 2013 to 2017

The oxide circuit has the following processes:

☐ Primary crushing.

☐ An optional secondary hybrid roll type crusher for the harder transitional and free-milling sulphide ores.

☐ Milling.

☐ Cyclone classification.

☐ Gravity concentration.

☐ Flash flotation.

☐ Carbon in leach.

☐ Tailings disposal.

18^th September 2018 Page 285

Kibali Gold Mine NI 43-101 Technical Report

The sulphide circuit has the following processes;

☐ Primary and secondary crushing.

☐ Milling.

☐ Cyclone classification.

☐ Gravity concentration.

☐ Flash flotation.

☐ Conventional flotation.

☐ Ultra-fine grinding of the concentrates.

☐ Pre-oxidation circuit.

☐ Pumpcell adsorption circuit to recover gold from the concentrates.

☐ Tailings disposal.

The loaded carbon from the pumpcell circuit, that is, from the concentrate leach and carbon in pulp together with carbon from whole-ore leach, are treated in independent elution circuits, followed by electro-winning of gold eluate.

Once the oxide, transition and free-milling ore sources have been depleted, the existing oxide plant can be converted to a parallel sulphide circuit, which will necessitate the expansion of the concentrate handling and pumpcell circuits. There are two flotation circuits already present in the plant.

Kibali further expanded the original fine-grind section in the 2017 sulphide expansion project by adding an additional four ultra-fine-grind mills, making eight in total.

Sulphide Crushing and Screening

Two primary jaw crushers (two Sandvik CJ815:200 kW, CSS:16 0 mm) are used targeting 1,300 tph and feeding two secondary crushers (two Sandvik CS660; 250 kW, CSS:45 mm) via a coarse ore stockpile (COS).

ROM sulphide ore, received from trucks, is treated in a primary crushing circuit comprising of a ROM bin, apron feeder, and primary jaw crusher (Sandvik CJ815) operated in open circuit at 1,300 tph target. This primary crushed product is then conveyed to a primary crushed stockpile or COS with a 5,000 t live capacity. The sulphide ore from underground has already been crushed underground and is also conveyed to this stockpile.

18^th September 2018 Page 286

Kibali Gold Mine NI 43-101 Technical Report

Apron feeders under the stockpile are used to combine the sulphide ores from the two sources before it is conveyed to the secondary crushing circuit that has two secondary cone crushers (Sandvik CS660) operating in parallel (running/standby) in open circuit to produce a crushed product stream with a P80 of 45 mm. The secondary circuit was commissioned in May 2014.

When sulphide ore is being treated, secondary crusher product is fed onto a fine ore stockpile (FOS) via a conveyor system. The FOS serves as a common mill stockpile to both the mills and has a live capacity of 11,700 t of sulphide ore to each mill. The mill is fed from the mill feed stockpile using apron feeders that feed directly onto the mill feed conveyor.

When oxide ore is being treated through its circuit, the primary crusher product (Sandvik CJ815) is fed directly to the mill feed conveyor and not via the secondary crushing circuit but may be subject to an in-line hybrid crushing stage, if deemed necessary, at least on one of the two parallel streams.

The design of the crushing circuit includes provision for the installation of a tertiary crushing circuit.

Oxide Crushing and Screening

ROM ore received from trucks is treated in a primary crushing circuit comprising of a ROM bin, an apron feeder, and a single toggle jaw crusher. This primary crushed product (at 450 tph) can either be diverted to the primary mill feed conveyor when oxide ore is treated, or alternatively conveyed to a common 5,000 t live primary crushed stockpile when sulphide ore is treated.

Sulphide Milling and Oxide Milling

A ball milling circuit comprising two Polysius ball mills, each operating independently in parallel, fitted with initially 7 MW, but later 8 MW motors, treat ore at a feed rate of 900 tph dry solids.

When treating oxide ore, the primary crusher product will feed directly onto the mill feed conveyor and into the milling circuit. When sulphide ore is treated, the mill will be fed from the mill feed stockpile.

Each ball mill is operated in closed circuit with a cyclone cluster used to produce a target grind of 80% passing 75 µm on sulphide and 80 µm on oxide. The mill feed consists of fresh crushed ore, a portion of the cyclone underflow, gravity concentrator scalping screen oversize and flash flotation cell high density tailings. Ground ore from the mill reports to the mill discharge sump where it combines with gravity concentrator tailings, flash flotation low density tailings and Gekko Inline Leach Reactor (ILR) tailings, before being pumped to the cyclone cluster.

When treating sulphide ore, the cyclone overflow is gravity fed to a rougher flotation circuit, whilst the overflow from the cyclones when treating oxide ore is directed to the CIL circuit. Cyclone underflow is split into three streams:

☐ Gravity concentration circuit.

18^th September 2018 Page 287

Kibali Gold Mine NI 43-101 Technical Report

☐ Flash flotation circuit.

☐ Remainder of the cyclone underflow is re-cycled to the mill feed.

Gravity concentrator tailings gravitates to the mill discharge sump while the concentrate reports to a batch ILR circuit.

The flash flotation cell produces a concentrate and a high-density tailings stream. The concentrate is directed to the concentrate handling circuit, alternatively it can be deposited into the feed of the gravity recovery pre-screening so as to enhance gravity recovery, while the high-density tailings are circulated back to the mill feed. Frother, collector, and promotor are added to the flash flotation cell for recovery of flash flotation concentrates. The required copper sulphate conditioner is added into the mill feed.

Flotation

Cyclone overflow from the primary milling circuit is routed to either the rougher flotation cells or bypasses the circuit to the rougher tailings tank before being pumped to the CIL circuit (if oxide or free-milling material is being treated).

Two separate parallel banks of six 70 m³ Outotec forced air rougher flotation cells in series are used for flotation; however, when one mill is processing oxide ore then only one bank is required. Frother, collector, activator, and promoter are added to the slurry stream of sulphide ore for the recovery of a flotation concentrate. Rougher flotation concentrates from the first three tankcells in each bank are pumped to the concentrate handling circuit where it can combine with the flash flotation concentrates. The concentrate from the last three tankcells in each bank is recycled to the flotation feed. The reason for this split of concentrates is to not only reduce the overall mass pull for the limited capacity downstream concentrate treatment circuits, but also maximise potential flotation recovery by ensuring the limited mass pull comprises the higher grade concentrate emanating from the first three cells whilst the lower grade concentrate from the last three cells is not lost as is simply recycled for further processing. The rougher flotation tailings stream is pumped to the flotation tailings thickener. Occasionally, when warranted, either for reasons of maintaining stability of parameters within the main CIL circuit, or in the event of a low flotation recovery, in other words a high flotation tails value, the float tail is routed for leaching in the main CIL circuit, thereby facilitating further recovery of the residual gold in the CIL circuit.

Intensive Leach Reactors (ILR’s)

Gravity concentrator concentrate is gravity fed into the reaction drum of the Gekko ILR (Inline Leach Reactor) from the feed cone. Sodium hydroxide, sodium cyanide, and oxygen are added in high concentrations to the reactor to put the gold into solution. At the completion of the leach the pregnant leach solution is pumped to the ILR electro-winning pregnant solution tank.

Oxygen is supplied from a 30 tpd oxygen plant being upgraded to a 40 tpd pressure swing adsorption (PSA) plant operated by Air Liquide.

Concentrate Handling and Ultra-Fine Grinding

18^th September 2018 Page 288

Kibali Gold Mine NI 43-101 Technical Report

Flotation and flash flotation concentrates, together with gold-room waste, report to the concentrate thickener. Thickener underflow is fed to the ultra-fine milling circuit. The ultra- fine grinding circuit (UFG) consist of eight VXP2500 FL Smidth (originally Deswik) ceramic bead mills in parallel, where circuit feed material of 80% passing 106 µm is treated to achieve a target grind of 80% passing 20 µm minimum, now typically 18 µm.

The ultra-fine milling products are pumped to the Pre-oxidation and pre-leach circuit followed by the pumpcell circuit.

Pre-Oxidation and Pre-leach

Slurry from the UFG circuit is fed to the Pump-cell circuit pre-oxidation and pre-leach circuit.

The pre-oxidation circuit consists of two agitated tanks operated either in parallel if both mill streams are treating refractory sulphide or in series if only a single stream on sulphide. Flexibility exists to operate as the ore demands. Each tank is fitted with four (two duty and two standby) Aachen REA450 reactors, through which slurry is circulated and contacted with oxygen. Both pre-oxidation tanks also have three oxygen sparge units. Lime and lead nitrate are dosed in both pre-oxidation tanks. Hydrogen Peroxide dosage is also available. The product stream from the second pre-oxidation tank overflows to an agitated pre-leach tank, fitted with one Aachen REA400 reactor with a dedicated Aachen reactor pump. The tank is also equipped with oxygen sparge units. Cyanide and lime are added to maintain the leach pH and hydrogen peroxide can also be dosed if required. A diesel dosing facility is also available to dose into the pre-leach tank and is aimed at reducing the effect of preg-robbing carbonaceous material in the slurry, though this is rarely if ever used. The pre-leach slurry product is pump fed to two 2100 m³ leach tanks for extended residence time in an Aachen assisted leach environment.

Pumpcells

The pre-oxygenated and pre-leached product stream overflows to eight 100 m³ Kemix Pumpcell tanks operated in series. Before the concentrate expansion project, there were only six tanks. Eight tanks have proved sufficient for twin-stream sulphide operation, both on account of the mass-pull reduction initiatives introduced to the flotation circuit, but also because the pumpcell circuit residue stream can be blended into the main CIL circuit to mop up any residual gold. The Pumpcell tanks are operated in a carousel mode, with counter current flow of carbon relative to slurry. Slurry is moved between tanks with MPS(P) screens which both transfer slurry to the next sequential tank and screens out the carbon which remains in the original tank. The carbon in a Pumpcell circuit always remains in the same tank but the position of the tank in the circuit is changed. One tank is isolated each day and the entire content of the tank is pumped to the elution circuit. It is then re-introduced to the circuit as the last tank and receives a fresh batch of carbon. This results in a loaded carbon batch size of 5 t per day. This process too is flexible where advantage can be taken of two tanks being harvested in a single day so as to completely fill an elution column.

Loaded Carbon and Tailings

18^th September 2018 Page 289

Kibali Gold Mine NI 43-101 Technical Report

Loaded carbon from the Pumpcell tank taken offline is pumped to a vibrating screen. The screen oversize reports to the elution circuit acid wash tank while the undersize is routed to CIL tank 3.

Tailings from the Pumpcell circuit exiting the last Pumpcell tank in the carousel are pumped to the CIL circuit.

CIL

Pre-oxidation

Oxide material, bypassing the flotation cells, is pumped to CIL tank No.1 for pre-oxidation via dedicated Aachen reactors

Lime is added to increase the slurry pH before cyanide is added. The leach pH level must be maintained at 10.5 or higher. Lead nitrate is also added. Hydrogen peroxide can also be used, if required.

The last two tanks in the CIL train have been dedicated to the extended pre-oxidation and pre-leaching of the concentrate stream prior to being routed to the pumpcell circuit.

CIL Tanks

Slurry and activated carbon flow counter current to each other through CIL tank numbers one to six. Slurry from the pre-oxidation tank flows by gravity to CIL tank 1. Pumpcell tailings also report to the CIL circuit with the flexibility of being treated in a selection of CIL tanks as deemed appropriate for the current ore feed blend and configuration, this after passing through the Pumpcell tailings samplers. The slurry in tank 1 is routed to tank 2, and thereafter each subsequent tank, using MPS(P) interstage screens. The MPS(P) screens only transfers the slurry while retaining the carbon in the tank.

Fresh/Re-gen carbon is added to the final tanks in the train and is pumped from each tank to the preceding tank in the sequence using carbon transfer pumps. A total of 12 t of carbon is transferred daily from each tank. Carbon and slurry is transferred by a carbon transfer pump from tank 1 to the loaded Carbon screen at the top of the elution columns where the carbon is screened out for elution. The screen underflow returns to CIL tank 1.

The following reagents are dosed in the CIL tanks- oxygen, cyanide, lime, lead nitrate, and hydrogen peroxide (used when oxygen is not available).

Detox Tanks

Four Detox tanks, each with two Aachen REA450 reactors, were designated to be used for cyanide destruction. Cyanide must be destroyed to below 50 ppm weak-acid dissociated (WAD) before CIL tailings can be disposed of in the lined tailings dam according to DRC legislation. The pulp in the detox circuit is typically contacted with activated carbon which serves as a catalyst for the oxidation of cyanide to cyanate. These tanks proved to be redundant both in terms of their ineffectiveness coupled with an alternative and superior method of controlling the cyanide

18^th September 2018 Page 290

Kibali Gold Mine NI 43-101 Technical Report

concentrations within the CIL circuit via a cyanide analyser and controller. Thus, they were transformed into normal CIL tanks, effectively lengthening the CIL train and hence extending the residence time, making up the end tanks of the main CIL circuit, well suited also for the dedicated treatment of extended pre-oxidation and pre-leaching of concentrate prior to being routed to the pumpcell circuit.

Elution

The CIL and Pumpcell carbon are batch treated in the AARL elution circuit separately through two identical circuits. The duplicate 12 t AARL columns share a common heater facility capable of running both columns simultaneously. The CIL carbon is to be treated in 12 t batches once every 24 hours, while carbon from the Pumpcell circuit will be treated in 10 t batches every 48 hours or more frequently if desired. The elution heaters are electric whilst the regeneration kilns are diesel fired.

Loaded carbon is collected in an elution circuit acid wash tank. Carbon that has been acid washed is then loaded into a 12 t AARL elution column by gravity. The carbon has a clear eluate solution (1% NaCN and 3% NaOH at 125°C) pumped through it that desorbs the gold from the carbon and puts into solution to form the loaded solution, which is then pumped to the elution electro-winning circuit feed tank/pregnant solution tank.

Barren carbon is removed from the elution column and reports to either of two carbon regeneration kilns.

Electro-Winning and Gold Room

Pregnant solution from the ILR circuit is circulated through a single electro-winning cell and steady head tank. Gold is deposited on the cathodes as sludge and the solution circulated until the desired barren gold concentration is achieved or 18 hours has elapsed.

Pregnant solutions from the CIL elution circuit and the Pumpcell elution circuit are treated in the same way except that there are six electro-winning cells operated in parallel for each stream.

Loaded cathodes are periodically removed from the cells, the gold sludge is washed off using a high-pressure washer and the washed mixture is then decanted. The gold sludge left behind is calcined in two electric calcination furnaces. The calcined sludge is then mixed with fluxes and loaded into an induction smelting furnace. After smelting, the furnace crucible contents are poured into cascading moulds to produce gold bullion and slag.

General

The plant has two distinct processing streams that are largely separate in most areas. Hence a loss area in one will not affect all of the output of gold in most loss scenarios. This includes the substations where, for example, each ball mill has its own substation. However, once the elution and carbon handling areas are reached the production of gold is concentrated into single work areas.

18^th September 2018 Page 291

Kibali Gold Mine NI 43-101 Technical Report

All equipment installed was new at the time of installation. The design of the plant considered the International Cyanide Code regulations. Kibali Goldmines is currently not a signatory, however endorses conformance with the International Cyanide Management Code.

In general, there is a satisfactory level of protection against collision damage in the structures that support elevated structures. Bypass arrangements for tanks mean that if one unit is offline, the process flow does not stop. There are only minor impacts on recovery efficiency if only one tank or flotation cell is offline.

The process plant has been built on an existing site. The site has been cut largely from the top of hills to accommodate the present structures. The crusher, bin and mill were built on the old leach pad area while the CIL tanks reside in the old pregnant solution pond area. Geotechnical boreholes and test pits were completed to design the terraces.

In summary, this is a new process plant that is meeting its throughput and recovery efficiency expectations. It can be considered as having matured to steady stated operations.

17.2 Processing Recovery

Overall, the actual process plant gold recovery in 2017 varied monthly from 80.2% to 85.6% (Figure 17-3 and Table 17-2). The average gold recovery in 2017 was 83.4%. Recovery for 2018 is expected to be 84%, increasing to 87% in 2019, and 89% in subsequent years based primarily on a shift away from a blended ore feed to one that will be dominated by the KCD and better recovery ores for Gorumbwa deposit.

Figure 17-3 Kibali Processing Plant Overall Gold Recovery in 2017

18^th September 2018 Page 292

Kibali Gold Mine NI 43-101 Technical Report

Table 17-2 Kibali Processing Plant Overall Gold Recovery in 2017 by Month

Item	Unit	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec	2017 Total
Tonnes Treated (Dry)	kt	646	603	672	632	644	578	607	619	613	677	644	683	7,618
Plant Head Grade	g/t	2.63	2.70	2.85	2.67	2.51	2.85	3.01	3.07	2.72	2.53	2.91	3.58	2.84
Recovery	%	80.2	82.0	81.1	81.9	85.6	85.1	82.0	84.3	84.1	83.1	84.6	85.6	83.4

The Kibali processing facility has largely seen improvements in its operational performance on a year by year basis, both in terms of throughput capacity, demonstrated previously in Figure 17-2. However, this performance extends to overall gold recovery as is evident by Figure 17-4 and Figure 17-5 depicting an improvement trend over the months of 2017 and previous years.

18^th September 2018 Page 293

Kibali Gold Mine NI 43-101 Technical Report

Figure 17-4 Kibali Plant Recovery

Figure 17-5 Kibali Plant Pumpcell Residue and Throughput

18^th September 2018 Page 294

Kibali Gold Mine NI 43-101 Technical Report

17.3 Production History

The Kibali processing plant has been operating since 2013. The production history is summarised in Table 17-3 and illustrated in Figure 17-6.

Table 17-3 Kibali Processing Plant Production History

	2013	2014	2015	2016	2017 to Nov
KCD Pit (kt)	531	5,573	4,724	1,814	1,186
KCD Pit Grade (g/t)	4.09	3.75	3.16	2.47	1.29
KCD Pit (oz)	70	671	480	144	49
KCD UG (kt)	-	70	709	1,369	1,561
KCD UG Grade (g/t)	-	3.87	4.72	4.76	5.29
KCD UG (koz)	-	9	108	209	265
Mofu Pit (kt)	-	83	84	-	-
Mofu Pit Grade (g/t)	-	4.91	4.55	-	-
Mofu Pit (koz)	-	13	12	-	-
Mengu Hill Pit (kt)	-	0	1,169	1,197	712
Mengu Hill Pit Grade	-	-	5.12	3.81	2.78
Mengu Hill Pit (koz)	-	-	193	147	64
Kombokolo Pit (kt)	-	-	-	288	438
Kombokolo Pit Grade (g/t)	-	-	-	2.69	2.91
Kombokolo Pit (koz)	-	-	-	25	41
Pakaka Pit (kt)	-	-	-	2,042	2,963
Pakaka Pit Grade (g/t)	-	-	-	2.34	2.23
Pakaka Pit (koz)	-	-	-	154	212
Rhino Pit (kt)	-	-	-	67	95
Rhino Pit Grade (g/t)	-	-	-	2.48	3.52
Rhino Pit (koz)	-	-	-	5	11
KCD Pit Pushback 3 (kt)	-	-	-	-	247
KCD Pit Pushback 3 Grade (g/t)	-	-	-	-	1.62
KCD Pit Pushback 3 (koz)	-	-	-	-	13
Total Feed (kt)	531	5,726	6,685	6,777	7,201
Total Feed Grade (g/t)	4.09	3.77	3.69	3.14	2.83
Total Feed (koz)	70	693	792	684	655

18^th September 2018 Page 295

Kibali Gold Mine NI 43-101 Technical Report

Figure 17-6 Kibali Plant Production History

17.4 Capital Projects and Plant Upgrade

The new cyanide tailings storage facility (CTSF2) (Figure 17-7) was extended to include a return water cyanide detoxification installation to remove the free cyanide from the return water to optimise the percentage of recycled process water. The height extension (1st lift) of the combined CTSF1 and CTSF2 tailings facilities commenced in Q4 2017. Lift 1 is planned to be completed in Q3 2018 with Lift 2 in 2020 and Lift 3 in 2024. The budget associated with this is $13.1 M. The extension allows for additional storage capacity.

18^th September 2018 Page 296

Kibali Gold Mine NI 43-101 Technical Report

Figure 17-7 Kibali Concentrate Storage Facility

The process plant expanded the concentrate handling facilities in anticipation of increased flotation mass pulls, especially when the mine became depleted of oxides and transitional ores resulting a full twin-stream feed of refractory sulphide ores. The expansion project addressed in a phased manner the following circuits:

☐ Rougher flotation circuit configuration modified to minimise mass pull but maximise recovery.

☐ Increase the ultra-fine grind capacity in Q1 2017, by installing an additional four vertical mills, bringing the total number to eight mills.

☐ The process plant expansion also includes an extended pre-oxidation and pre-leach circuit through reconfiguring of the pre-oxidation circuit to cater for the increased flows expected, but also incorporating the last one, two or three tanks in the CIL train.

☐ Increase of the pumpcell capacity from six to eight tanks.

☐ The budget for this project was $12.8 M.

☐ The increase in sulphide ores in 2017 is clearly seen in Figure 17-8.

18^th September 2018 Page 297

Kibali Gold Mine NI 43-101 Technical Report

Figure 17-8 Kibali Processing Weekly Twin Stream Sulphide Proportion

17.5 Processing Costs

Operating Costs (OPEX)

The 2016 and 2017 actual processing operating cost can be found in Table 17-4.

Table 17-4 Actual Process and Plant Engineering Operating Costs for 2016 and 2017

Cost	Units	2016 Actual	2017 Actual
Fixed Cost
Consultants	$’000	203	347
Contractors - Assays	$’000	1544	1,588
Contractors - Oxygen	$’000	1735	1,634
Equipment Hire	$’000	3299	2,852
General Costs	$’000	6655	6,903
Gold Refining	$’000	3233	3,917
Labour	$’000	4896	5,734
Stores - Other	$’000	391	1
Total Fixed	$’000	21,956	22,976
Tonnes Processed	kt	7296	7,619
Total Fixed	$/t	3.01	3.02
Variable Costs
Power	$/t	3.90	4.49
Reagents - Cyanide	$/t	3.10	2.80
Reagents - Lime	$/t	1.20	0.60
Good Issues - Caustic Soda	$/t	0.67	0.58
Good Issues - Activated Carbon	$/t	0.16	0.09
Reagents - Other	$/t	2.01	2.02
Stores - Grinding Media	$/t	0.81	0.96
Stores - Liners	$/t	0.41	0.43
Stores - Screens and Panels	$/t	0.01	0.10
Total Variable	$/t	12.27	12.07
Total $/ t	$/t	15.28	15.09
Plant Engineering	$/t	3.79	3.62
Combined Plant & Engineering	$/t	19.07	18.71

LOM processing costs have been budgeted at $17.24/t (which included plant engineering cost). The actual costs for 2017 were $18.71/t, with the key improvements being over the LOM being

18^th September 2018 Page 298

Kibali Gold Mine NI 43-101 Technical Report

the two hydropower stations that have been brought online in 2018 and 2019, which will drop power cost by approximately $1.30/t. Further to this, cyanide consumption has been optimised to a level of $2.50/t for the LOM (2017 levels being $2.80/t) as the plant operations are stabilised.

18^th September 2018 Page 299

Kibali Gold Mine NI 43-101 Technical Report

18 Project Infrastructure

18.1 Mine Roads

The Kibali Mine is located in the NE of the DRC. Access to site by road is via Uganda and the Ugandan border town of Arua. The road from the Ugandan border at Arua has been upgraded by the company to accommodate the Project and on-going operations traffic. Maintenance of this road is carried out by the company.

The local road infrastructure was developed during the exploration drilling programmes and upgraded during the construction of the mine. Internal roads provide access to various infrastructure areas, including roads to the TSF, Explosives Storage, Land Fill Site, Mine Villages, Central Mine Offices, Shaft Collar Area, Open Pit Mining Central Operations Area, general mining operations areas, new exploration areas, various water boreholes, and overhead line routes.

All roads are constructed by layered rock/gravel/laterite varying in specification according to traffic expectations.

18.2 Supply Chain

Since the Project’s inception, the majority of Kibali’s imports are shipped into the port of Mombasa, Kenya, and thereafter trucked through the Northern Corridor road route that links Mombasa to the landlocked countries in Eastern and Central Africa. The, cargo initially moves through Kenya and Uganda into Eastern DRC (Kibali). Up to the Uganda / DRC border, the trucks use a two-way tarmac road considered to be the main route from the port of Mombasa to East and Central Africa. The final 200 km of the trip from the DRC border to Kibali is on laterite roads.

The primary ports for mining spares and consumables are Durban and Antwerp. Reagents, such as cyanide, steel balls, peroxide, hydrochloric acid, and other flotation reagents are shipped from a variety of different ports worldwide. The shipping terms for the mining consumables and reagents are typically Ex-Works or Free On Board and Cost, Insurance and Freight respectively.

The costs associated with 20 ft and 40 ft containers, for both sea-freight and inland transport (Mombasa to Kibali mine site), are calculated on a cost-plus basis. This is a fully transparent exercise with shipping/freight invoices being sent through for verification.

18^th September 2018 Page 300

Kibali Gold Mine NI 43-101 Technical Report

Estimated port to port transit times for Kibali’s most frequent sailings:

☐ South Africa = 10 days

☐ Europe = 35 days

☐ China = 45 days

☐ USA = 65 days

Procurement for Kibali Goldmines is carried out by the Supply Chain partner, namely, Tradecorp Logistics.

18.3 Surface Water Management

Kibali lies within the northern tropical climatic region of the DRC. The area has a distinct rainy and almost dry season. The rainy season extends from March to November and the dry season from December to mid-February.

The Kibali River dominates the drainage of the Project area and flows along the southern boundary of the Project area. The Nzoro River flows into the Kibali River approximately 30 km downstream of the Kibali site. Numerous springs exist in the area and the spring flows remain near constant throughout the dry season.

The significant sources of water that can affect the operations include rainfall directly into the open pits, rainfall surface run-off and groundwater entering the pits from the surrounding rock masses.

Surface run-off is high, due to high intensity rainfall events and an undulating landscape. A system of bund walls and dewatering trenches has been established prior to mining of each of the pits, which prevents inflow of surface water to the pit areas. The network of drainage channels is used to discharge water intercepted by the perimeter drains to the Kibali River via a series of settling ponds.

All the deposits are characterised by the presence of a near-surface groundwater table with the potential for high groundwater into the pits and the possible impacts of ingress of groundwater are investigated prior to mining and during the mining activities. Dewatering well systems are installed for all pits to lower the groundwater level prior to mining.

The rainfall that falls within the pit perimeter is directed out of the pit, if this is possible, particularly in the upper levels. The water that cannot be directed outwards flows to the sump at the pit bottom from where it is pumped.

Figure 18-1 presents an overview of the Kibali Water Management Plan.

18^th September 2018 Page 301

Kibali Gold Mine NI 43-101 Technical Report

18.4 Water Supply

Raw water is collected and stored in the raw water dam (RWD), which has a storage capacity of 9,500 m³. The primary source of raw water is rain and spring water catchments with a dam top-up from a borehole system and final backup from the Kibali River.

The processing plant requires 46,000 m³ of water per day. The primary sources of plant process water are as follows:

☐ FTSF return water.

☐ CTSF return water.

☐ Concentrate thickener overflow.

☐ Flotation tailings thickener overflow.

☐ Storm water.

The plant process water circuit consists of a 25 m diameter process water clarifier and process water dam with a capacity of 5,000 m³.

The operational camp has an independent water purification plant and storage facility.

18.5 Tailings Facilities

Two TSFs exist at Kibali; one for the cyanide containing CIL tails and one for the sulphide flotation tails. The CIL tails contain small amounts of cyanide and must be contained in a plastic lined dam. The flotation tails contain no harmful substances and therefore the dam is not lined.

The cyanide containing TSFs comprise CTSF1 and CTSF2 for the CIL tails and the FTSF is dedicated to flotation tails.

Epoch Resources (Pty) Ltd (Epoch) designed the facility and has prepared a LOM Strategy for the CTSF and FTSF. The facilities are managed by Fraser Alexander Tailings, as the Company’s engineer.

A large volume of the tailings generated by the plant will be used for underground backfill in future. It is estimated that up to 50% of the flotation tailings will be used for paste backfill.

The CTSF currently consists of two fully HDPE-lined basins; CTSF1 and CTSF2 that have a continuous surrounding embankment and share a common internal wall. CTSF1 covers an area of approximately 64 ha and has reached full capacity. Tailings are no longer being depositing into this basin. A second compartment, CTSF2, has been constructed and is now in use. An area of approximately 45 ha of CTSF2 is currently covered with tailings.

18^th September 2018 Page 302

18th September 2018 Page 303

Kibali Gold Mine NI 43-101 Technical Report

Minimum freeboards for CTSF1 is 1.5 m and for CTSF2 is 2.5 m vertical freeboard and 1 m beach freeboard.

CTSF Phase 3 is presently envisaged as being self-raising from the final elevation of tailings for first two phases. Based on the current LOM plan, deposition into the Phase 3 facility would commence in Q4 2019 and continue until the end of the expected mine life.

The FTSF will require a LOM capacity of approximately of approximately 57.8 Mt tonnes with Phase 1 providing storage for approximately 8.2 Mt. Tailings are being also deposited into the FTSF with approximately 60 ha of Phase 1 of the facility currently covered with tailings. Total footprint is 123 ha.

Phase 1 is currently in operation as an unlined valley impoundment, formed behind an embankment that traverses the valley. An unlined Return Water Dam (RWD) captures and stores return water from the FTSF. Phase 2 consists of a raise of the existing compacted wall and construction of other smaller compacted wall around the facility to create a full containment to a specified height. Phase 3 is envisaged as a full ring dyke impoundment that will operate under self-raising conditions.

An unlined catchment dam captures water from the area of the TSFs.

18.6 Power Supply

Thermal and Hydropower Stations

Since there is no grid power available in the region, Kibali needs to be self-sustaining and indeed possesses considerable thermal power generation capacity to do so. Diesel generated power comes from three banks of on-site high-speed diesel generators, each bank consisting of twelve x 1500 kVA, 400V CAT 3512B generators. In order to mitigate the running costs of this facility, feasibility studies have demonstrated the benefits associated with the installation of several potential hydropower plants. To date two such plants have been installed with a third nearing completion.

These are as follows:

☐ Nzoro 2 Four x 5.5 MW turbines - Total installed 22 MW

☐ Ambarau Two x 5.3 MW turbines - Total installed 11 MW

☐ Azambi Two x 5.3 MW turbines - Total installed 11 MW

Nzoro 1 already existed but is of low capacity at well less than 1 MW. It was refurbished and represents a historical legacy comprising equipment dating from the 1930s. This power is dedicated to local communities.

The Nzoro 2 hydropower station was optimised during 2015 and reached its design power supply (22 MW) by the start of 2016. Commissioning of the second new hydropower station, Ambarau, was achieved in 2017 and the completion of a third station, Azambi, scheduled for 2018 is

18^th September 2018 Page 304

Kibali Gold Mine NI 43-101 Technical Report

expected to deliver hydropower in line with the 2012 feasibility study, along with a consequent drop in power costs.

The long-term power supply strategy for the operation is aimed at generating the maximum amount of power from hydro sources. Diesel generators will remain available as back up and as a spinning reserve for peak loads from the shaft hoist. This has a marked effect on reducing the unit power operating costs. Wet seasons with high river flows allows for more beneficial hydro operating conditions, however the beneficial effect is still seen in the lower rainfall months. This effect is evident in Figure 18-2 which shows the power supply mix to the end of 2017.

Figure 18-2 Kibali Electrical Supply Mix

Once Azambi has been completed, total installed hydroelectric power capacity will be 42 MW, which should cover most of the mine power demand. Full power demand at full production is anticipated to be between 39 MW and 43 MW. It is currently expected that once the hydro strategy is fully implemented, diesel generators will supply only 7 MW, with the remainder being provided by the hydrostations.

Therefore, the system has a potential capacity of 44 MW of Hydropower (at peak) and 32 MW of thermal Generation. Actual hydro generation capacity is season dependent:

☐ Maximum Capacity (32 + 44) MW.

☐ Minimum Capacity (32 + 10) MW.

The load demand of the mine is not constant, and the average power consumption is approximately 40 MW.

Electrical power at 66 kV is supplied by the hydropower stations connected to a main grid supply.

The hydro-generated power is reticulated to the site by means of 66 kV overhead lines from the Hydropower Plants to a switchyard located at the mine. The voltage is be stepped down from 66 kV to 11 kV, feeding the 11 kV consumer substation.

18^th September 2018 Page 305

Kibali Gold Mine NI 43-101 Technical Report

Diesel generated power supplies power to the mine at 400V, which is stepped up to 11kV for distribution.

18.7 Site Infrastructure

Operational Camp (Village)

The operational camp provides accommodation for single and married staff and incorporates all the required facilities in terms of accommodation, ablution, catering, and messing facilities.

The camp comprises two villages to accommodate the mine employees; a large single status camp near the mine operations and a married-quarters camp that was opened in 2015.

A single kitchen and dining room is provided for the people resident in the camp. A further kitchen and dining area is available at the social club that could be used if the camp kitchen was destroyed. Each of the major contractors operates their own camp and kitchen facilities.

Offices, Stores, and Workshops

A central administration area office complex accommodates senior and administrative personnel as well as discipline functions not located specifically in the process plant or mine operations offices.

The plant area includes the necessary buildings for the operations personnel related to the process operation including a gate house, control room containing the plant server and SCADA equipment, engineering room, UPS rooms, engineering offices, laboratory including carbon room, metallurgical laboratory, wet laboratory, bullion room, balance room, environmental laboratory, receiving area, sample preparation and grade control preparation, and a maintenance workshop and offices.

The Central Mine Facilities Area is located adjacent to the processing plant and includes large stores facilities the spares and

18^th September 2018 Page 306

Kibali Gold Mine NI 43-101 Technical Report

engineering consumables for mining, processing, and general operations.

There are four large buildings to hold most of the stores stock, most of which is spares for machinery, and the remainder is consumables, such as personal protective equipment. There is sufficient undercover space for the spares and consumables.

The buildings are all steel framed and clad with steel sheeting. Floors are reinforced concrete.

The shaft collar area provides an office building, change house, security gate house, and a workshop for the underground mining operation.

The open pit mining central operations area includes a large workshop for the maintenance of the mining fleet, an office building, a change house, and a security gate house.

Emergency Response and Medical facilities

There are two mine rescue teams on-site with a total of seventeen active members of which ten are on-site at all times. These people all have positions on the mine with the Mine Rescue being an additional responsibility.

Emergency situations will be communicated by radio on a dedicated channel. A stench gas system is available.

A fire truck and trailer is available for the rescue teams.

Medical staff on site includes two doctors, six nurses, and laboratory technicians. There are three ambulances on site and four first aid rooms together with a health clinic.

The nearest hospital with good facilities is in Kampala. In the event of a need for medivac, arrangements with the air charter company would made.

The company runs a malaria prevention programme involving bush clearance and spraying, and a campaign to improve awareness. This has resulted in a significant reduction in malaria cases. The programme is ongoing. In addition, there is a continued HIV Aids campaign including voluntary counselling and testing.

Fuel Storage

The fuel storage installation includes three separate fuel farms.

Daily consumption is approximately 180,000 l during the wet season and 200,000 l during the dry season. Approximately 65 to 70% of the consumption is used by the diesel generators at the thermal power station, 20% is used by mining and the remaining 10% is general use.

The largest fuel farm is located in the Central Mine Facilities Area. The main fuel farm for the mine has three one million litre tanks and six 100,000 l tanks, giving a total storage capacity of 3.6 Ml. Diesel is filtered before it is pumped into the main tanks and after it leaves the slave tanks.

18^th September 2018 Page 307

Kibali Gold Mine NI 43-101 Technical Report

Extensive fire protection is provided for the main fuel farm and includes a series of foam generators located around the perimeter of the containment bund and also cooling rings on the tanks. The water for these fire protection systems is supplied from two dedicated tanks and two fire pumps located at the process plant.

Two other fuel farms have been built at the open pit and underground operations and have a capacity of 1,200 m3 each with similar dispensing facilities.

Airstrip

Access by air to Kibali involves a commercial flight to Entebbe in Uganda followed by a charter flight to Doko airport, situated on the mine property. The Doko airstrip was upgraded by the company and is equipped with runway lights and precision approach path indicator lights.

Charter flights to site are arranged by the company on a regular schedule at frequencies dictated by operational requirements.

18.8 Communication and Information Technology

The mine wide voice and data backbone with satellite fibre optic link(s) provides cellular for voice and internet connections via wireless LAN. Voice communication is supplemented by two-way radio.

Fibre optics on overhead lines provide for communication between the various operations sites.

18.9 Security

There is comprehensive security infrastructure at the site, with controlled to the operations. The Security Manager reports directly to the Kibali Goldmines General Manager.

The Kibali mine property is surrounded with a high fence and a security access road running along the perimeter.

The plant area is fenced with security at the main gate and additional electronic access systems and security at higher values areas within the plant.

The spares and materials storage sites are fenced and access gates are kept locked and access controlled by security staff.

18^th September 2018 Page 308

Kibali Gold Mine NI 43-101 Technical Report

19 Market Studies and Contracts

19.1 Revenue, Tax and Royalty

Financial evaluation of all Ore Reserves uses a gold price of $1,000/oz and with the exception of Ore Reserves for the KCD pit which uses a $1,100/oz gold price optimised pit design. All other open pit Ore Reserves are estimated within pit designs which are based on a gold price of $1,000/oz. This is in line with Randgold’s corporate guidelines. Gold price sensitivities were run for all the pits and the decision on a higher price for the KCD is discussed in more detail in Section 15.

Financial evaluation and cut-off grade calculation for the Kibali underground Ore Reserves has been based on a gold price of $1,000/oz. This same value has been used for all previous Kibali underground Ore Reserves estimate from 31st December 2011 onwards.

Royalties payable to the DRC government remain unchanged from completion of the feasibility in 2012. A total royalty payable to the DRC government of 3.5% of gold revenue inclusive of 1% shipment fees was used for the open pit Ore Reserve estimate.

Kibali currently pays income tax at a rate of 30% to the DRC government. Due to accelerated depreciation charged on capital expenditure, a tax shield has been built up, meaning taxation payment will only commence in 2024.

19.2 Marketing

Gold doré produced at the mine site is shipped from site under secured conditions and sold under agreement to Rand Refinery in South Africa. Under the agreement, Kibali Goldmines receives the ruling gold price on the day after dispatch, less refining and freight costs, for the gold content of the doré gold. Kibali Goldmines has an agreement to sell all gold production to only one customer. The “customer” is chosen periodically on a tender basis from a selected pool of accredited refineries and international banks to ensure competitive refining and freight costs. Gold mines do not compete to sell their product given that the price is not controlled by the producers.

19.3 Contracts

It is Kibali Goldmines strategy to outsource mining activities to contractors and, in all instances, the contracts are such that the equipment can be purchased by the company at the end of the contract period at its depreciated price or should the contractor default at a predetermined pricing mechanism. Prior to start-up all major mining contractors are requested to tender and the most appropriate tender is accepted thereby ensuring that the best competitive current pricing is achieved. Care is taken at the time of finalising contracts to ensure that the rise and fall formula

18^th September 2018 Page 309

Kibali Gold Mine NI 43-101 Technical Report

is totally representative of the build-up of the quoted price per unit. At the time of award prices quoted are compared to benchmark prices of other owner miner operations.

The contract mining costs are dependent on when tenders are issued as the price of major equipment varies dependant on demand as well as the cost of finance. Rise and fall can be negatively affected by currency fluctuations as well as price squeeze due to scarcity.

The mine produces doré bars which are sent to an accredited gold refinery for refining. Refining prices are subject to fluctuations in the cost of transport as well as insurance costs. Other contracts that are put in place include assay facilities, oxygen supply, catering services, fuel supply, explosive supply, and security.

18^th September 2018 Page 310

Kibali Gold Mine NI 43-101 Technical Report

20 Environmental Studies, Permitting, And Social or Community Impact

20.1 Environmental Considerations

The Project consists of multiple deposits, being mined by a combination of several open pits and an underground mine First gold was produced in 2013.

Waste rock is disposed of close to the open pits, and oxide and sulphide ore is trucked to and processed at a central plant using crushing, grinding (including an ultra-fine grind), gravity, flotation and CIL followed by smelting to produce doré bullion. Tailings are disposed of in two tailings facilities, the first unlined facility is used to store the flotation tails (FTSF), while the second lined facility holds the concentrate tails (CTSF), which are acid producing and which also contain cyanide residues and the higher arsenic containing materials. CTSF facility is composed of two lined TSFs; CTSF1 and CTSF2; and are due to be incorporated during the life of the operation to form one facility. A portion of the flotation tailings are used for paste backfill in the underground KCD operation, which will be at full capacity in 2018.

Two hydro-power plants have been built; Nzoro 2 and Ambarau which significantly reduce dependence on diesel fired power plants and thus also reduce greenhouse gas emissions (GHG). In addition the existing with Nzoro 1 hydrostation has been refurbished by Kibali Goldmines and is exclusively used to provide power to the local community. The Azambi hydropower plant will be completed in 2018 with first power is scheduled for 2018.

Significant resettlement caused by the impact of the KCD mining operations and involving 17,000 people was undertaken during the development of the Project, with the establishment of the Kokiza Host village, to allow for future mining activities. Economic displacement (loss of land, crops, trees, and access to natural resources) has also taken place across the 3,150 ha Exclusion Zone of the Project.

Environmental Assessment and Permitting

An independent Environmental and Social Impact Assessment (ESIA) for the Kibali mine was completed as part of the Kibali Goldmines feasibility study during the course of 2010 and 2011 with approval from the authorities received in 2011.

The ESIA, as well as complying with the DRC legislation and the DRC Mining Code (2002), was developed according to the Equator Principles and the International Finance Corporation (IFC) Performance Standards, forming part of the IFC’s sustainability Framework.

A separate ESIA was completed in June 2011 for a new hydropower station, Nzoro 2 and refurbishment of the existing Nzoro hydropower station (Nzoro 1) adjacent to the Kibali and Nzoro rivers respectively. This includes the upgrading of the existing power lines from the Nzoro 1 station, construction of new power lines from Nzoro 2 as well as the construction of a diversion

18^th September 2018 Page 311

Kibali Gold Mine NI 43-101 Technical Report

canal off the Nzoro River to the Nzoro 2 station. An ESIA was also completed in 2012 for the Ambarau and Azambi Hydropower plants located on the Kibali River.

The Project is largely governed by the DRC Mining Code (2002) and the Mining Regulations (2013) which contain various provisions including those around Environmental Impact Assessment (EIA) and environmental management, public consultation, and compensation for loss of access to land.

Key permits, licenses and compliance acquired since the Project acquisition include: an Environmental Adjustment Plan, an import and export licence under Kibali Goldmines, permit for the construction of infrastructure at Kokiza, authorisation to import explosives, demolition permit, authorisation to resettle people, authorisation for exhumation (so that graves can be relocated out of the mining zone), title deeds for all people resettled in Kokiza and authorisation for the construction of four hydropower Stations.

Under the DRC Mining Code (2002) mining operations, which were in existence at the time the DRC Mining Code (2002) came into force, must be covered by an Environmental Adjustment Plan (EAP) approved by the Direction de Protection de l’Environnement Minier (DPEM). The purpose of the EAP is to give an overview of the environmental condition of the areas covered by the relevant mining title under which such operations are conducted and to describe any measures that have been or will be taken to protect the environment. In practice this plan will also need to cover what is normally required in an Environmental Impact Study (EIS) and an Environmental Management Plan (EMP). Schedule IX (Contents of EIS and EMP) of the Mining Regulations sets out the contents of the EIS and the EMP and provides detail regarding specific management measures and standards that are required.

Public consultation of the Project has been achieved in accordance with articles 126 and 127 Schedule IX of the Mining Regulation and the social description of the Project site has been completed in accordance with Article 38 of the same schedule.

Mitigation and rehabilitation measures at Project closure have been included in the EAP. These measures are quoted in the EAP in accordance with Chapter VII Schedule IX of the Mining Regulation Articles 95 and 123.

The assessment of the environmental impact studies and the plans of environmental management, presented in the EAP was made by the Standing Committee of Evaluation (CPE) comprising 14 members and directed by the Director of the Mining Environmental Protection, the manager of the DPEM. The EAP was assessed and approved by the CPE, required under Articles 455 and 456 of the Mining Regulations.

The EAP was reviewed and approved by an inter-ministerial committee in accordance with Article 455 of the Mining Regulation. Copies of the EAP were submitted to the Mining Registry Office as requested under Articles 69, 92, 103, 154 of the DRC Mining Code (2002) and Article 454 of the Mining Regulations.

18^th September 2018 Page 312

Kibali Gold Mine NI 43-101 Technical Report

During 2014 and 2015 the ESIA was updated to consolidate the many separate ESIAs which had been prepared for separate elements of the Project, and to update the collective impacts of the Project. It was approved by the DRC government in 2016. In addition, the update of the ESIA was required in compliance with the national mining environmental legal requirements where a five-year ESIA update is mandatory. This is to ensure that the operation stays up to date on current best practices and continually manages its impacts. It allows for a re-examination of the management processes and responsibilities and assists the mine in continuously managing its environmental and social impacts.

All environmental permits are in place for the Kibali processing plant, open pits and underground operations as well as the hydropower stations, and a permit register forms part of the EMP. Kibali Goldmines continues with the rehabilitation of disturbed areas (although rehabilitation of the Mengu Hill and Rhino pits planned for 2017 was halted to allow exploration of the underground mining potential of these pits), monitoring of water discharge qualities, standard waste disposal procedures, waste dump rehabilitation and adequate and standard tailings storage facility.

Environmental Management and Monitoring

The Environmental and Social departments report to the mine General Manager of the mine with functional reporting to the Group Community and Environmental Officer, who provides technical support when and if needed. Each department develops its own budget and programmes in consultation with the Group Community and Environmental Officer, who undertakes regular audits and inspections to ensure the site is compliant. Kibali is unique in the Randgold Resources Group in having a separate Social and Community department, which reflects the close proximity of significant centres of population and close ties with the resettled communities in the area.

An EMP is in place which covers all aspects of the operation, and the Kibali operations are ISO 14001:2015 compliant and independently audited to continuously improve environmental management. External audits are also carried out for compliance with the International Cyanide Management Code (ICMC). Gaps identified in the last ICMC audit, carried out in September 2017, relating to transportation, handling and storage, operations, decommissioning, worker safety, emergency response and community dialogue are being addressed.

Environmental performance objectives are established by senior management and communicated within the operations. All the strategic inputs are formalised within this and close tracking is carried out through the EMS.

Routine environmental monitoring takes place across the site, including dust deposition, noise, various sampling for arsenic and WAD cyanide (which can be analysed on site) including TSF seepage water and tails streams and sample collection of drinking water, groundwater, surface water and TSF borehole water, which are sent to an external laboratory. Energy use is monitored, and records show that 50% of the energy consumed during 2017 was provided by the two hydropower dams. Environmental incidents are noted in a register which forms part of the EMS; causes and responses identified, and incidents closed out. A total of 24 minor incidents were reported in 2017.

18^th September 2018 Page 313

Kibali Gold Mine NI 43-101 Technical Report

Waste is segregated and managed either by landfilling, incineration, or storage/recycling. New opportunities are being sought for reusing or recycling the stored materials.

A Water Quality data review was carried out in February 2016, which examined the data base of water quality results collected to date and which had the following conclusions:

☐ The data for most monitoring points are incomplete with large gaps in data and some points not being monitored regularly;

☐ The groundwater monitoring point in and around the mine camp is constantly showing parameters to be above the recommended drinking water guidelines. This water should be treated before used for drinking or domestic water;

☐ NO₃ and NO₂ levels in all monitoring points have increased in the last year to levels above the recommended guideline concentrations;

☐ TDS, pH and SO₄ levels are stable and within the guideline ranges;

☐ Arsenic and CN is not monitored in the groundwater points; and

☐ Arsenic has increased to higher levels in surface water points in the last year.

The deposits have variable amounts of arsenic; these are identified during mining and any high arsenic ore is stockpiled and blended with low arsenic ore through the plant to provide an even throughput. Arsenic has also been identified within the mining and process streams during routine water quality monitoring and led to a more detailed assessment of the FTSF seepage which so far, has not revealed any concentrations that exceed permitted standards.

A comprehensive water balance model has recently been developed for the site, which models flows, inputs and losses across the complex site, including the open pits and underground workings, plant, TSF, water management structures, offices, camp, and treatment facilities. The dynamic model also identifies river water use, dewatering water, discharges, gains, and losses identifies volumes of potential savings/recycling opportunities which would reduce abstraction rates from the river. Some work is required for full implementation of the dynamic model, such as the installation of flow meters on a number of pipelines, and this is scheduled for 2018.

Geochemistry

Geochemical characterisation of the waste rock samples was undertaken using industry standard techniques which included testing of carbonate; however, the carbonate was found to be iron rich. The presence of the iron rich carbonate complicated interpretation of the acid neutralising capacity of the material. A 25% reduction in the measured neutralising capacity was therefore applied to allow for the uncertainty. The waste material was found to have moderate to high acid neutralising capacity for the majority of lithologies tested.

Acid base accounting showed that five percent of the samples were classified as being potentially acid forming. This is considered a very low proportion of the total waste. All potentially acid forming samples had sulphide contents of 0.8% or higher and showed little correlation to gold grade or depth. Net acid generation (NAG) testing of the waste was generally consistent with the results of the acid base accounting which employed the reduced acid neutralising capacity values.

18^th September 2018 Page 314

Kibali Gold Mine NI 43-101 Technical Report

The waste was found to present a moderate to high salinity risk with the majority of the material tested poorly suited for use in the outer facing of the waste dumps.

Waste samples were found to be enriched in a limited number of elements with arsenic and antimony being the most highly enriched elements. The elemental enrichment showed a poor correlation to sulphide content or gold grade. Water extraction testing found that the majority of the enriched elements had low solubilities under certain neutral pH conditions. Arsenic, molybdenum, barium, cobalt, zinc, nickel, and selenium were found to be above the guideline water quality values in one or more samples, with arsenic and molybdenum the most commonly elevated elements. Twelve samples had water extraction (leachate) arsenic concentrations above the guideline water quality values and eight had water extraction (leachate) molybdenum concentrations above the guideline water quality values. The leach potential of the material did not show a strong correlation to the total concentration of the elements contained within the waste.

Waste rock is used to build various infrastructural platforms on site, while the remainder is stockpiled on surface or deposited in stopes as backfill. ARD has not been detected at the site despite the uncertainties around the predictions of Net Neutralisation Potential.

Biodiversity

The Project area lies within the Northern Congolian Forest Savanna Mosaic ecoregion that includes the northernmost savanna woodlands in Africa (White, 1983). This narrow transition zone marks an abrupt habitat discontinuity between the extensive Congolian rain forests to the south and the Sudanian / Sahelian grasslands to the north. This forest-savanna mosaic represents the eastern half of the Guineo-Congolian / Sudanian phytogeographical regional transition zone.

The Northern Congolian Forest Savanna Mosaic ecoregion is typified by a combination of gallery forest, woodland and secondary grassland and is controlled by annual precipitation, duration of water stress and the severity of dry-season fires and human activity.

The original vegetation of the Project area has been largely transformed through human activity. Shifting agriculture when practiced at low intensity permits natural secondary succession of rain forest species. However, as the local population has increased, traditional agriculture has become less successful. Population pressure particularly in the last 100 years has shortened the average fallow period from 20 years to as little as three, reducing the ability of soils to regenerate and regain fertility. The increased human activity has reduced tree densities and created extensive sparsely wooded grasslands.

During baseline studies in 2010, three survey sites were selected as being representative of the vegetation existing within the Project area. Based on the results of the flora surveys, the vegetation of the Project area has been broadly characterised into moist savannas and bush fallow, gallery forest, agricultural areas, and exotic plantations.

Three plant species of conservation significance were recorded within the Project area.

18^th September 2018 Page 315

Kibali Gold Mine NI 43-101 Technical Report

Albizia (Albizia ferruginea) is considered to be of conservation significance. It is listed as being Vulnerable by the International Union for Conservation of Nature (IUCN). It is described as a widespread and often common timber species, which has suffered heavy exploitation. The major threat is from timber exploitation. The IUCN descriptive code for Albizia ferruginea is ‘VU A1cd’. This tree was encountered in gallery forest and secondary forest only.

Guarea cedrata (also called Light Bossé or Scented Guarea) is a species of tree in the Meliaceae family; it is exploited in the timber industry and has therefore listed as threatened IUCN ‘Vulnerable A1c’. This tree species was only encountered in the gallery forest vegetation type.

Pterygota bequaertii is a species of flowering plant in the Sterculiaceae family; it is listed as being Vulnerable by the IUCN ‘VU A1cd’. This tree species was only encountered in the gallery forest vegetation type. Forest, woodland, and secondary grassland integrate in patterns controlled by rainfall, drought, fires, and human activity. Gallery forests are the dominant habitat type in the Northern Congolian Forest Savanna Mosaic. The ratio of forest to savanna in this area is likely to have fluctuated considerably over time, driven by long-term climatic changes, the vagaries of rainfall and dry season length as well as by shifts in human disturbance patterns. The ecoregion thus provides a unique set of habitats and resources that supports a moderate level of diversity, including many species with broad distributions in tropical Africa. This ecoregion has received relatively little scientific attention, making broad habitat assessment and characterisation difficult.

No faunal species of international conservation significance were identified during the surveys, although these were limited due to the security situation at the time of the surveys. Despite human pressure, both the gallery forest and the moist savannah are in fairly good condition and are home to several habitat specific species.

It is obvious that the mosaic of habitat is fundamental to the continued high species diversity in the region. Many species will breed and/or roost in the forest patches or savannah and forage in the agricultural areas. The gallery forest along the Kibali River functions as an important ecological corridor while the river itself is utilised by several species of birds.

Hunting is widespread in the study area. Evidence of this was found in the gallery forest and the moist savannah. Despite this hunting pressure, several species which were recorded during the brief survey are favoured by bush meat hunters and their continued presence in the area is considered a good indication that hunting has not been totally destructive.

The reported presence of elephant approximately eight kilometres to the north of the study sites puts the remoteness of the site into context and is a good indicator that the area is connected to the wilder more remote areas in the Province. It is also an indication that it is probable that many more sensitive species may enter the study area on occasion.

An ecological integrity state assessment was undertaken on the local Kibali River and Durba Dams during May 2007 and follow up studies in 2010 and 2011, in order to characterise the baseline and identify the impacts, if any, associated with the recent SOKIMO (Kilo-Moto) operation and associated artisanal mining activities. An additional component considered for this study included an assessment of any potential ecological implications associated with the

18^th September 2018 Page 316

Kibali Gold Mine NI 43-101 Technical Report

draining of the Durba Lakes into the Kibali River, which needed to take place during the construction phase of the Project. The conservation of biodiversity as well as the sustainability of natural resource management were considered and addressed for the Project in line with IFC Performance Standards. Water quality in both the Kibali and Nzoro Rivers was found to be good and should not have a limiting effect on aquatic biota. Habitat integrity in both rivers remained largely unmodified with existing impacts primarily limited to the riparian zone.

Additional aquatic biomonitoring was carried out during the dry season in 2017, in streams within the fence perimeter and surrounding rivers: new species were identified in many of the tributaries and one area of improvement was identified downstream of the FTSF. Only invertebrates were present in this stream.

Additional biodiversity monitoring is ongoing, such as the use of camera traps to detect fauna within the concession. The Biodiversity Management Plan (BMP) is being updated to reflect additional information on biodiversity which has been collected. The mine site lies around 65 km south of the Garamba National Park, which lies on the border with South Sudan. A partnership with the Park has been established and environmental representatives from Kibali, together with local students undertake educational visits to the Park. Other assistance and support is provided, including providing scientific survey information (for example aquatic biodiversity surveys), supporting the collaring and tracking of elephants, supplying fuel to anti-poaching teams, supporting the Kordofan giraffe programme and building road infrastructure such as bridges within the park to assist the rapid response of ranger patrols. This partnership provides a wider strategic support for game protection from poachers from the north, and connections with local enforcement networks.

During 2017, 12,230 trees were planted around the site, for rehabilitation and erosion control purposes. An additional 1,450 seedlings were selected from the Azambi HEP area for further reforestation. A partnership is in place with local farmers to supply tree seedlings to the Kibali nursery.

Mine Rehabilitation and Closure

Mine rehabilitation will be an on-going programme designed to restore the physical, chemical, and biological quality or potential of air, land and water regimes disturbed by mining to a state acceptable to the regulators and to post-mining land users. Current rehabilitation opportunities are limited due to the stage of the Project. While some pits are worked out and waste rock dumps are inactive, these are being assessed for potential future underground operations so will not be rehabilitated.

The activities associated with mine closure are designed to prevent or minimise adverse long-term environmental impacts, and to create a self-sustaining natural ecosystem or alternate land use based on an agreed set of objectives. The objective of mine closure is to obtain legal (government) and community agreement that the condition of the closed operation meets the requirements of those entities, whereupon the company’s legal liability is terminated.

18^th September 2018 Page 317

Kibali Gold Mine NI 43-101 Technical Report

A preliminary closure plan was developed which is regularly updated and refined to reflect changes in mine development and operational planning, as well as the environmental and social conditions and circumstances. Records of the mine works will also be maintained as part of the post-closure plan.

The preliminary framework addresses the following:

☐ The regulatory framework for mine closure, providing the legislative requirements to be considered with closure planning;

☐ The overall closure goal and associated objectives for the various mine sites; and

☐ The broad closure measures for relevant components of the mine sites.

This framework aims to identify and list all closure elements, with their associated rehabilitation and closure objectives in such a manner so as to facilitate systematic and progressive measurable closure throughout the life of the operation. By planning for closure at an early stage, the calculated rehabilitation and closure liability will be reduced throughout the life of mine.

Mine closure costs are updated each year, with increases or decreases in disturbed areas noted and costed; the current cost for rehabilitation and closure of the mine according to the calculation model is $32 M Kibali’s environmental liability.

Allowance has been made for the shaping of the open pit edges and WRDs to a safe and sustainable angle. Rehabilitation of the Run of Mine (ROM) Pads, demolition and management of physical infrastructure, creation of a free-draining topography, replacement of soil, re-vegetation, and general surface rehabilitation of all the disturbed areas within Kibali has also been calculated. It has been assumed that infrastructure (i.e. brick buildings) at the airport, mine camp and mine offices will be left for the community after mine closure, and the extension areas (i.e. mainly contractor laydown areas) will be rehabilitated by the contractors as per the contractor’s agreement with Kibali. The cost of demolishing and rehabilitating these areas has therefore been excluded from the closure cost assessment.

A contingency of 10% has been included to allow for areas which may have been undervalued or overlooked. A 12% allowance has been included for project management fees, which in the absence of any DRC guidelines are based on the South African Department of Mineral Resources (DMR) closure guidelines. These fees account for the costs required to manage the closure and rehabilitation.

The total includes costs for surface and groundwater monitoring for five years after mine closure, monitoring and maintaining re-vegetated areas for three years after mine closure, bi-annual aquatic biomonitoring for five years post-closure, hydro-carbon clean-up, and cyanide decontamination.

20.2 Social Considerations

Employment and Procurement

18^th September 2018 Page 318

Kibali Gold Mine NI 43-101 Technical Report

Management of Human Resource is key for Kibali Goldmines in ensuring high productivity and efficiencies of the workforce. Kibali Goldmines conforms to the labour laws of the DRC. This includes:

☐ Salary and remuneration Scale of employees

☐ Job classification and competencies

☐ Annual leave system

☐ Expat to national workforce ratios

☐ Unions representative

☐ Employee code of conduct and Disciplinary measures and

☐ Mine Level Agreement (MLA).

Kibali Goldmines seeks to employ highly skilled employees for its various disciplines. With the internal in reach program implemented, a platform where both employer and employee work to ensure the employee engagement concept is alive.

Priority on employment is to fill positions with DRC nationals where available. In the absence of a national skill, expatriate employees with special skills are brought in with the primary aim of training nationals to the required level to take over within a given time frame. Where a proper Personnel Development Plan is in place for a better succession plan and skills transfer. Sourcing of national skills involves looking at the nearby community within the Permit before moving to other regions of the country.

Training programs both in-house and outside the mine are periodically organised by Kibali Goldmines for experts and consultants to up the skills and equip employees with adequate skills and knowledge.

HR procedures include an employee’s recognition process and an employee grievance mechanism. The recognition process was not well understood by employees and resulted in 23 grievances which have been addressed through meetings with Unions and departments to ensure parties understand the process. The transition from contract mining to owner mining is planned to take place in mid-2018.

2017 figures show that the workforce was made up of 83% contractors, of whom 92% were nationals, and 17% employees, of whom 88% were nationals. Overall local employees number 4,917 (92%), out of a total of 5,377 employer and contractor posts.

The underground mining operations contribute to extended life-of-mine, employment of local Congolese and the growth of the DRC economy.

There is also a policy of promoting local procurement which also covers contractors. Where possible, goods and services should be procured locally. This includes produce from the various agribusinesses (eggs, pork, maize) which is purchased for use in the mine canteens.

Resettlement

18^th September 2018 Page 319

Kibali Gold Mine NI 43-101 Technical Report

As a result of the construction of the Project, it was necessary to resettle approximately 17,000 people, from 4,000 households, from the immediate Project area, referred to as the Exclusion Zone. The Project also displaced around 134 items of community infrastructure, including 13 communal agricultural projects, five communal business/commercial facilities, 12 education facilities, 19 health facilities, nine recreational/community facilities, 39 religious facilities and 41 water sources.

The Exclusion Zone for the Project comprises approximately 3,150 ha. Most of this land has been turned over to the Project for the mine and associated infrastructure. The land was utilised primarily for residential sites, agricultural activities and ASM mining activities. Consequently, there was a loss of land and associated resources for those communities residing on the land and/or making use of the resources found on it. The loss of this area of land also places greater demand and pressure on other agricultural land, trees, and other natural resources. Due to the prevalence of ASM in the area, agriculture is however, not practised as intensively as other rural areas.

In addition, the Project drained the two Durba dams in order to enable pre-construction activities to commence. The loss of this water source was not significant since the water is not potable and utilised by villagers primarily for washing purposes e.g. clothing and motorcycles, and also ASM miners, i.e. washing of gold ore. Since the communities surrounding the two dams were resettled and the ASM activity ceased, only temporary mitigation measures were implemented.

Kibali Goldmines is committed to the following resettlement principles:

☐ To apply whichever of the two sets of guidelines (DRC legislation and IFC PS 5) is most favourable to the Project Affected Peoples (PAPs).

☐ To follow a resettlement and compensation process that will leave PAPs in the same or better off position than before the Project intervention.

☐ To follow in-kind compensation where possible and limit cash compensation as far as possible, especially where the affected community’s livelihoods are at stake.

If cash compensation is insisted upon by the PAPs, the Company will support this in compliance with the demands of best resettlement practise but will endeavour to ensure that it is used responsibly where possible. For example, if people decide to build their primary infrastructure themselves, payment will be made in instalments and full payment will only be made upon completion of construction to ensure the structure is rebuilt.

Compensation was decided on through extensive consultation with the Resettlement Working Group (RWG). The RWG is the primary consultative forum for the purposes of the generation of the Kibali Resettlement Action Plan (RAP) and as such was established for the purpose of advising the government of the DRC and Kibali on the orderly and equitable resettlement of those people affected by Kibali. All primary stakeholders are represented on the RWG. Budget and monitoring plans were put in place to ensure effective execution of the resettlement action plan.

Basic services including water, electricity and transport options are very limited in the Project area. The RAP included construction of water, energy, and road infrastructure. Guidance was

18^th September 2018 Page 320

Kibali Gold Mine NI 43-101 Technical Report

provided by Congolese town planners, as well as the RWG, for a town plan outlining the development of the host site that improved the provision of basic services and social infrastructure whilst still being maintainable, considering the overall remoteness of the area.

The major RAP was initiated in 2012 and completed during 2013. Kibali Goldmines maintains regular liaison with local stakeholders and community leaders. Employment of local skills is a key priority for the operation. An independent RAP completion audit was carried out in 2013 when nine of the fourteen affected villages had been relocated to the resettlement site, Kokiza. The objectives of the audit were to identify any social/community issues or risks associated with the Project not currently being addressed, focusing on the following four Project components: Mine site, Resettlement host site, Doko-Aru Road upgrade; and Nzoro hydropower station and power line. Also assessed was the implementation of the RAP against commitments set out in the RAP documentation, which was developed in line with International Finance Corporation’s (IFC) Performance Standard 5.

It must be noted that since the audit the above improvements and recommendations have been implemented.

Gorumbwa RAP was initiated in 2016 to allow for the future mining of the Gorumbwa Pit. The resettlement was applied to households situated within the mining perimeter, and was based on compensation and construction of a replacement property. Additionally, eight boreholes have been drilled at Kokiza and handed over to the community as well as all community infrastructures. There were some threats by Gorumbwa communities to engage in public protest due to delay in compensation payments, however, this was abated by the engagement of the social team with RAP affected parties.

Stakeholder Engagement

Stakeholder engagement activities, community development projects and local economic development initiatives contribute to the maintenance and strengthening of Kibali Goldmines Social License to Operate (SLTO).

In 2017, Kibali Goldmines reinforced its relationship with the Provincial Government as a number of meetings were organised between the two institutions regarding the collaboration on various topics: Master Plan for the Development of the Province of Haut-Uélé, the evaluation of the engagement act, creation of the steering committee for the Construction of the General Hospital of Watsa and others.

Kibali Goldmines has maintained strong relations with the community through continuous stakeholder engagement which includes regular meetings with a range of stakeholders and regular radio broadcasts targeting key issues pertinent to the community. A full programme of stakeholder engagement meetings takes place with the Community Development Fund, GM monthly and quarterly meetings, media, Civil Society, Watsa-Faradje-Aru Youth, ASM bi-monthly meetings, one-on-one meetings, and radio broadcasts. Recent development initiatives include investment in agriculture and livestock projects including maize, pork, poultry, and fish as well as

18^th September 2018 Page 321

Kibali Gold Mine NI 43-101 Technical Report

undertaking feasibility studies for cocoa, bananas, and palm oil plantations. The company is a significant employer to members of the local communities.

The collaboration between Kibali Goldmines and the community keeps improving as discussions are more constructive and participative. The relationship between Kibali Goldmines and the community is growing and becoming more productive. No major issue has been reported in 2017 and all complaints and grievances brought by the community were resolved peacefully. In 2018 the mine will continue to focus on targeting major issues faced by the community in: education, health, agriculture, general infrastructures, and various campaigns (for example road security, HIV, malaria)

The Social and Community department has a Stakeholder Engagement Plan which is updated every year, also a Social Licence Strategy and a Community Development Plan.

Community Development/Corporate Social Responsibility

2017 projects included youth apprentice training in masonry, carpentry and welding, work on progressing a large-scale water distribution project in Durba and investment for the improvement of access to potable water in Aru. A total of $1.95 M was spent on community development projects in 2017.

Existing projects include egg, pork, and animal feed production. Much of the produce forms a significant portion of local food supply to the Kibali catering contractors. In 2017, the proportion of local against imported goods ranged between 70 and 100% local.

Community healthcare initiatives include support to Watsa hospital (development of a business plan), Central Hospital Kibali (CHK) hygiene and housekeeping (CHK has contracted a local company to perform daily housekeeping for three months with possibility of renewing it whilst progress is monitored on a daily basis by Kibali Goldmines doctors) and support to the Watsa Orphanage, where a noticeable improvement in the health status of orphans and housekeeping (in the court yard and dormitory) has occurred due to the involvement of the clinic.

Sport and cultural activities are a component of CSR activities, and activities included the handover of sport equipment to FC Ouragan and motorcyclist associations.

A capacity building program has provided training for more than 2,500 teachers in the Watsa Territory, 40 women in agri-processing, provided by a South African consultant, and Lectures and students from the Isiro Technical College visited the mine and community-funded initiatives. In addition, certificates were presented to 121 youths trained in masonry, welding, and carpentry.

20.3 Artisanal and Small-Scale Mining

Despite the potential boost to the local economy, the Project has resulted in a loss of economic activity and livelihood for some people, particularly those benefiting from the artisanal and small-scale mining (ASM) economy within the exclusion zone, including those people employed by SOKIMO. According to reports, the Kibali sector had an estimated 34 ASM workings, with

18^th September 2018 Page 322

Kibali Gold Mine NI 43-101 Technical Report

approximately 13 workings in the EZ. The closure of ASM sites affected those sites within the EZ only.

Although the artisanal miners are bound, in terms of the DRC legislation, to make way for the Project as the legally entrenched industrial mining enterprise, they had the potential to be a formidable obstacle to orderly resettlement and development. The closure of ASM pits not only affects people directly involved in ASM activity, but it also has a knock-on effect on small businesses and farmers who sell goods to the ASM miners in the village. This impact was most experienced by villages that relied on ASM pits that were closed by the Project, but who resided some distance away from the exclusion zone and consequently not ideally positioned to benefit from the economic opportunities of the Project.

For many of the resettled communities, ASM represented a significant proportion of their household income, and households spent more time mining than farming. Livelihood replacement with non-mining activities has therefore been a focus of the RAP activities.

Nonetheless, artisanal mining remains a concern in the Kibali Permit area and Kibali Goldmines is working with the provincial authorities to eliminate ASM within the Permit. Activities have been observed to have expanded and are close to the R26 road, which constitutes a threat to the main access road for the province, and other road blockages have occurred, but this has been quickly resolved with police intervention.

The provincial authorities have signed a number of orders for a moratorium and the programme for the cessation of ASM in compliance with the mining code.

In Makoro, miners dug an underground mine that has crossed the Aru - Doko national road with the risk of landslide or the destruction of the road. This could impact access for people and supplies to the city of Durba and the mine. The provincial authorities were informed, inspected the workings, and suspended the mining activity and closed the workings. However, other workings are still active in Faradje territory. There are plans in place to develop a cessation strategy and engage with the Haut Uélé governor for his full involvement in the cessation process, as well as sensitising the local community and local chiefs to continue ASM activities in the ‘corridors’ identified for ASM by the government.

Influx

As with all large-scale commercial mining developments, it can be expected that there will be an influx of people into the area in search of jobs or to take advantage of the economic growth during construction and operations. In the case of the Project, there has been huge growth in the local community which now has an established infrastructure and services such as banking. The region’s isolation from other parts of the DRC, particularly to the south, where the Congo Basin extends to the western edge of the Rift Valley and the borders with Rwanda and Burundi, has meant that this settlement has become a regional hub which attracts opportunistic job seekers as well as artisanal miners from other parts of the DRC, as well as from other parts of Africa. The existing ASM activities and past LRA activities in the Project area have resulted in social ills already being present in the Project area.

18^th September 2018 Page 323

Kibali Gold Mine NI 43-101 Technical Report

21 Capital and Operating Costs

21.1 Capital Costs

Basis of Estimate

The Kibali Mine is an on-going combined open pit and underground mining operation with the necessary facilities, equipment, and manpower in place to produce gold.

The basis for the combined LOM plan is the Proved and Probable Ore Reserves estimate described in Section 15.

In the Qualified Person’s opinion, the open pit and underground LOM and cost estimates have been completed in sufficient detail to be satisfied that economic extraction of the Proved and Probable Ore Reserves is justified.

The majority of the capital cost estimates contained in this report are based on quantities generated from the open pit and underground development requirements and data provided by Kibali Goldmines.

Capital expenditure over the remaining LOM is estimated to be $370.6 M, made up from the following allocation of costs.

Construction and Projects

Construction and Projects capital cost predominantly include the cost for the completion of Azambi Hydropower Plant and the associated infrastructure.

Ongoing Capital

Ongoing Capital costs include the overhaul and replacement costs for underground mobile mining equipment and services and administration vehicles.

Underground Capital Development and Drilling

This category covers the cost to completion of the northern leg of haulage level drive and on-going LOM capital waste development. Capital development costs are based on a calculated average cost per metre for development including development of declines, decline stockpiles, ventilation drives, level access drives and long hole ventilation raises.

Pre-Production Capital

Pre-production capital covers open pit waste stripping.

A summary of capital requirements anticipated over the LOM is summarised in Table 21-1.

18^th September 2018 Page 324

Kibali Gold Mine NI 43-101 Technical Report

Table 21-1 LOM Capital Expenditure

Description	Value     ($ ‘M)
Construction and Projects	22
Ongoing Capital	99
Underground Capital Development and Drilling	203
Pre-production capital	9.3
Exploration	5.4
Rehabilitation/Mine Closure	32
Total LOM Capital Expenditure	371

21.2 Operating Costs

Basis of Estimate

The open pit mining operation is a contractor-run operation. To date the underground mine operation has also been contractor operated, however during 2018, the operation will begin moving to an Owner-operated mine.

Kibali Goldmines maintains detailed operating cost records that provide an excellent basis for estimating the future operating costs.

The basis for the combined LOM plan is the Proved and Probable Ore Reserves estimate described in Section 15.

In the Qualified Person’s opinion, the open pit and underground LOM and cost estimates have been completed in sufficient detail to be satisfied that economic extraction of the Proved and Probable Ore Reserves is justified.

Costs used for the open pit optimisations were derived from the Mining Contractor’s pricing of the open pit LOM schedule. Owners cost were also added.

Labour costs for national employees were based on actual costs. Local labour laws regarding hours of work etc. were also considered and overtime costs included.

During 2017 costs for processing and G&A were updated based on actuals adjusted for the latest forward estimates, production profiles and manning levels.

Customs duties, taxes, charges and logistically costs are included.

LOM Operating Costs

Unit costs used to estimate LOM operating costs are summarised in Table 21-2. The annual fluctuation in production levels is relatively low, such that the effect of fixed versus variable expenses is minimised.

18^th September 2018 Page 325

Kibali Gold Mine NI 43-101 Technical Report

For the underground mine, operating costs are derived from 2017 actual costs for Kibali. LOM costs have been adjusted to reflect operational changes including the move from contract mining to Owner operations from 2018.

Table 21-2 provides the unit cost inputs to the LOM operating costs.

Table 21-2 LOM Operating Unit Costs

Activity	Units	Value
Open Pit Mining - Kibali	$/t mined	3.27
Open Pit Mining - Kibali	$/ore tonne mined	21.62
Underground Mining	$/t mined	34.46
Underground Mining	$/ore tonne mined	35.88
Stockpile Movement	$/t milled	0.30
Processing	$/t milled	17.20
G&A	$/t milled	7.78
Mining Total	$/t milled	30.40
Total LOM Net OPEX	$/t milled	55.58

Kibali has used the unit costs to estimate LOM operating costs. Operating costs for the LOM plan are shown in Table 21-3.

Table 21-3 LOM Operating Total Costs

Description	LOM Total Cost ($ ‘M)
Open Pit Mining	479
Underground Mining	1,617
Processing	1,189
Stockpile	21
G&A	537
Total Operating Cost	3,842

The LOM has been prepared on the basis that the underground production mining activities will be transition to Owner operated mining during 2018. It is assumed that current contract prices will remain unchanged for mining activities performed by a contractor such as open pit mining and the underground development and production.

Cost inputs have been priced in real Q4 2017 dollars, without any allowance for inflation or consideration for changes in foreign exchange rates.

18^th September 2018 Page 326

Kibali Gold Mine NI 43-101 Technical Report

22 Economic Analysis

This section is not required as the property is currently in production, Kibali Goldmines is a producing issuer, and there is no material expansion of current production. Kibali Goldmines has verified the economic viability of the Ore Reserves via cash flow modelling, using the inputs discussed in this report.

18^th September 2018 Page 327

Kibali Gold Mine NI 43-101 Technical Report

23 Adjacent Properties

23.1 Kibali South

Kibali South Exploration Permit is located 2.5 km SW of the KCD pit in an exclusion zone surrounded by the Kibali Exploitation Permit. Kibali South is owned by SOKIMO, however, Vector Resources reported on 26^th February 2018 that they have undertaken technical and financial due diligence on the property for the purpose of joint venture with SOKIMO. The final details of this process have not been disclosed at the date of this report. Kibali South was previously owned by Kibali Goldmines and was transferred to SOKIMO in December 2012.

The mineralisation is an up-plunge projection of mineralisation below the KCD 9000 lodes, and is refractory in nature.

An historic non-JORC/NI 43-101 compliant Mineral Resource was estimated at 28 Mt at 1.63 g/t Au for 1.5 Moz.

23.2 Zani Kodo

The Zani Kodo exploration project is not adjacent property as it 60 km from the Kibali Permit. not immediately adjacent to the Kibali licence. It is, however, a significant regional deposit which is located on comparable geological setting to Kibali. The property is owned by ASA Resource Group PLC (80%) and SOKIMO (20%). In September a Mineral Resource estimate of was reported that contained 6.3 Mt at 3.25 g/t for 659 koz of Indicated Resources, and 3.0 Moz of Inferred Resources.

Zani Kodo is a shear related mineralisation style deposit with a total strike length of 5 km. The main mineralised zone is between 20 m and 30 m thick, continues 900 m down dip (open), and continues some 600 m along strike to the NNW where it appears to pinch out or be fault displaced. The main zone consists of silicification in sheared greenstone, and banded iron formation with gold associated with sulphides

Test work on samples from the Kodo Main mineralisation showed the ore to be non-refractory, allowing 90% recovery of contained gold. (2015 Annual report).

18^th September 2018 Page 328

Kibali Gold Mine NI 43-101 Technical Report

24 Other Relevant Data and Information

No additional information or explanation is necessary to make this Technical Report understandable and not misleading.

24.1 Country Risk

The following is taken from the Annual Report on Form 20-F 2017 submitted to the US Securities and Exchange Commission.

Randgold are subject to risks associated with operating the Kibali mine in the DRC. The Kibali mine is located in the north-east region of the DRC and is subject to various levels of political, economic and other risks and uncertainties associated with operating in the DRC. Some of these risks include political and economic instability, high rates of inflation, severely limited infrastructure, lack of law enforcement, labour unrest, and war and civil conflict. In addition, the Kibali mine is subject to the risks inherent in operating in any foreign jurisdiction including changes in government policy, restrictions on foreign exchange, changes in taxation policies, and renegotiation or nullification of existing concessions, licenses, permits and contracts.

The DRC is an impoverished country with physical and institutional infrastructure that is in a poor condition. It is in transition from a largely state-controlled economy to one based on free market principles, and from a non-democratic political system with a centralised ethnic power base to one based on more democratic principles. There can be no assurance that these changes will be effected or that the achievement of these objectives will not have material adverse consequences for the Kibali mine.

Any changes in mining or investment policies or shifts in political attitude in the DRC may adversely affect operations and/or profitability of the Kibali mine. Operations may be affected in varying degrees by government regulations with respect to, but not limited to, restrictions on production, price controls, export controls, currency remittance, income taxes, foreign investment, maintenance of claims, environmental legislation, land use, land claims of local people, water use and mine safety. These changes may impact the profitability and viability of the Kibali mine.

Moreover, the northeast region of the DRC has undergone civil unrest and instability that could have an impact on political, social, or economic conditions in the DRC generally. There has been turmoil in the Eastern DRC, to the south of Kibali, following the defeat of the M23 rebel group in late 2013. In March 2016, certain open pits at Kibali were overrun by artisanal miners, the resolution of which required the involvement of the State security forces, which temporarily disrupted the operation of these pits. In late 2016, political tensions arose stemming from a constitutional crisis surrounding the presidency. Delays in the presidential elections, now scheduled for December 2018, have led to protests and increased tensions in the country. The failure to secure a peaceful transition of power could lead to armed conflict and pose a significant risk to the country’s stability. A sufficient level of stability and effective national and local

18^th September 2018 Page 329

Kibali Gold Mine NI 43-101 Technical Report

administration must be maintained in order for Randgold to continue to operate the Kibali mine. The impact of unrest and instability on political, social, or economic conditions in the DRC could result in the impairment of the exploration, development and operations at the Kibali mine.

Goods are supplied to Randgold’s operations in the DRC primarily by road through Kenya and Uganda, which at times have been disrupted by geopolitical issues. Any present or future policy changes in the countries in which Randgold operates, or through which Randgold are supplied, may in some way have a significant effect on Randgold’s operations and interests.

24.2 DRC Mining Code Review

The DRC Mining Code (2002) and Regulations have been amended with an updated Mining Code which came into force on 9th March 2018 (DRC Mining Code(2018)) and the related amended Mining Regulations which came into force on 8th June 2018.

The following changes have been made to the DRC Mining Code (2002) that could have an impact on Kibali:

☐ Royalty charges are to be increased from 2.5% to 3.5%. This increases royalty charges over the LOM by an estimated $94.5 M, which would not materially impact the LOM profitability.

☐ Various increases in import and other duties from 4% to 7% depending on consumable type, which would not materially impact the LOM profitability.

☐ A super-tax profit has been promulgated based on the feasibility study prepared at the time the approval was given for the building of the Project. Accordingly, such a tax would only apply if the average annual gold price was in excess of $2,000/oz.

The exact impact, if any, of the changes will only be fully known once the 2018 Mining Code and related regulations are clarified and implemented in full.

Going forward the DRC Mining Code (2018) envisages a stability period for the tax, customs and exchange control regime of five years from the date on which the DRC Mining Code (2018) came into force and further provides that a number of the taxes shall be applied in accordance with the applicable substantive law.

Kibali Jersey Limited, the holding company of Kibali, the shareholders of Kibali Jersey Limited and Kibali Gold Mines SA, are considering all options to protect their vested rights under the DRC Mining Code and to enforce the additional state guarantees previously received, including preparations for international arbitration. In addition, engagement with the DRC government is ongoing, with the aim of exploring alternative solutions, which could be mutually acceptable to both parties. This includes the application of Article 220 of the DRC Mining Code (2018), which affords benefits to mining companies such as Kibali, operating in landlocked infrastructurally challenged provinces. If Article 220 were applied to Kibali, any advantages granted would mitigate any impact of the implementation of the DRC Mining Code (2018).

18^th September 2018 Page 330

Kibali Gold Mine NI 43-101 Technical Report

The QP notes that the mining code is in the process of transition and the current proposed changes do not have an impact on the stated Mineral Resource or Ore Reserves at the gold prices used for Ore Reserve ($1,000/oz), Mineral Resource ($1,500/oz), or the current gold prices of $1,200/oz (August 2018).

18^th September 2018 Page 331

Kibali Gold Mine NI 43-101 Technical Report

25 Interpretation and Conclusions

25.1 Geology and Mineral Resources

Kibali has documented standard procedures for the drilling, logging, and sampling processes, which meet industry standards. The geological and mineralisation modelling at Kibali is based on visibly identifiable geological contacts, which ensure a geologically robust interpretation can be developed.

Kibali has a quality control program in place to ensure the accuracy and precision of the assay results from the analytical laboratory. Checks conducted on the quality control database indicated that the results are of acceptable precision and accuracy for use in Mineral Resource estimation.

Geological models and subsequent Mineral Resource estimations have evolved and improved with each successive model update from added data within both open pit and underground. Significant grade control drill programs, and mapping of exposures in mine developments have been completed to increase the confidence in the resulting Mineral Resources and Ore Reserves. This was demonstrated in 2017 as this was the first time that Proved Ore Reserves have been disclosed for the underground mine.

In the QP’s opinion, the Kibali Mineral Resources top capping, domaining and estimation approach are appropriate, using industry accepted methods. Furthermore, the constraint of underground Mineral Resource reporting to use optimised mineable stope shapes has been deemed to reflect good to world best practice by external project audits. The QP considers the Mineral Resources at Kibali are appropriately estimated and classified.

The QP is not aware of any environmental, permitting, legal, title, socioeconomic, marketing, metallurgical, fiscal, or other relevant factors, that could materially affect the Mineral Resource estimate.

The strategic focus for the Project exploration is to prioritise additions of resources and reserves at open pit satellite projects to extend the life of open pit operations. Additionally, underground resource definition down plunge extension drilling is currently focussed on the target areas above the base level of the shaft thereby decreasing potential development requirements to access such areas for production.

25.2 Mining and Ore Reserves

The open pit mining operations at Kibali consists of multiple open pits. The open pits are being operated by a mining contractor and a down-the-hole blasting service will be provided by an appropriate blasting contractor. Opportunities exist with the Inferred Mineral Resource within the current pits that can be upgraded and converted to Ore Reserve with drilling. A reduction of open pit production is scheduled from 2023 and the end of current open pit mine life is estimated at year 2026 based on current Ore Reserves.

18^th September 2018 Page 332

Kibali Gold Mine NI 43-101 Technical Report

The Kibali KCD underground mine is designed to extract the KCD deposit directly beneath the KCD pit. A 50 m crown pillar separates the pit bottom from the top of the underground mine. The underground mine is a long hole stoping operation planned to produce at a rate of 3.6 Mtpa for 10 years. The majority of the underground mine infrastructure is already in place. A vertical production shaft is scheduled for full commissioning during 2018 following commissioning of the materials handling system. In 2018, the production will move to the majority of ore being hoisted up the shaft, however, throughout the underground LOM the decline to surface will be used to haul ore from some of the shallower zones and to supplement the shaft haulage.

The LOM has a long tail of declining production over a further nine years. The schedule will be progressively optimised as underground exploration of down plunge extensions progresses to extend the period of 3.6 Mtpa production rate.

Randgold, as the owner operator of Kibali Goldmines, has significant experience in other mining operations within Africa and these production rates, modification factors, and costs are benchmarked against other African operations to ensure they are suitable, taking into account the increased relative cost of fuel and labour within the DRC.

The current Ore Reserves for Kibali support a total mine life of 15 years at near full mill capacity, nine years of open pit operations, and 15 years of underground mining. LOM gold production averages approximately 542 koz per year.

The QP considers the modelled recoveries for all ore sources and process plant combined process and engineering unit costs, used within the Mineral Resource and Ore Reserve process to be acceptable.

The QP is not aware of any environmental, legal, title, socioeconomic, marketing, mining, metallurgical, infrastructure, permitting, fiscal, or other relevant factors, that could materially affect the Ore Reserve estimate.

25.3 Processing

Extensive metallurgical testwork campaigns have been completed across all mineral deposits in Kibali that form part of the declared Ore Reserve. These have consistently demonstrated two distinct behavioural patterns the first of which exhibits free-milling characteristics suitable for gold extraction by a conventional CIL metallurgical process and the second of which exhibits a degree of refractoriness, where straight cyanidation returns gold dissolutions considered to be too low for optimal plant operation due to the presence of occluded gold particles within sulphide minerals. It has been demonstrated that a finer grind will expose a portion of this additional gold for leaching so that the recovery is enhanced to economically acceptable levels.

The Kibali process plant operational risks are materially reduced as a function of the two separate process streams and independent milling circuits. The process plant has demonstrated excellent improvements in throughput capability, even performing beyond the design capacity at 7.2 Mtpa at consistent recovery performance.

18^th September 2018 Page 333

Kibali Gold Mine NI 43-101 Technical Report

The ore feed plan is blended using both KCD underground ore plus ore sourced from satellite open pits at Kibali in order to provide a stable feed grade blend. The Kibali feed plan utilises geometallurgical models that estimate the arsenic content within potentially arsenic bearing mineral deposits such that any ore with high arsenic contents is stockpiled separately and blended into the CIL process route to ensure that discharge is directed to the lined CTSF and discharge levels are below the environmental requirements.

Operational risks in the plant are reduced as a result of having two separate process streams and independent milling circuits.

The QP consider the modelled recoveries for all ore sources and the process plant and engineering unit costs applied to the Mineral Resource and Ore Reserve process to be acceptable.

25.4 Environment and Social

Kibali has a maturing environmental and social management plan and an accredited ISO14001:2015 Environmental Management System (EMS) in place which addresses current operational needs and can readily be adapted to meet future activities. Mine closure costs are reviewed and revised annually in line with good international industry practice.

All permits are in place and an Environmental Adjustment Plan has been approved by the DPEM.

The mine prioritises local employment and in 2017 achieved 88% Congolese employment across Kibali Goldmines employees and 92% across contractor workforces.

Stakeholder engagement is ongoing, and all senior management are involved in regular meetings with the community.

Two significant resettlement campaigns have taken place, one in 2012/2013 and one in 2016/2017. Ongoing monitoring of affected households to ensure that their livelihoods, often previously based on artisanal mining, are not adversely affected by the resettlement, will be ongoing. Economic displacement has also been significant across the area.

Artisanal mining remains a concern in the Kibali Permit area and the mine is working with provincial authorities to eliminate ASM within the Permit.

Kibali Goldmines continues to invest in community development initiatives, focussing on potable water supplies, primary school education, health care education, investment in medical clinics and local economic development projects.

The mine is a significant employer to members of the local communities. The mining operations contribute to extended life-of-mine, employment of local Congolese and the growth of the DRC economy. Kibali Goldmines policy is to promote nationals to manage the project.

18^th September 2018 Page 334

Kibali Gold Mine NI 43-101 Technical Report

The QP considers the extent of all environmental liabilities, to which the property is subject, to have been appropriately met.

25.5 Ownership and DRC Mining Code

The Kibali operation conforms to the DRC mining code and regulations. The next renewal date for the Permits are on 5^th November 2029 and 6th March 2030 and the current life of mine plan for the Kibali Ore Reserves extends beyond these dates.

The DRC mining code (2002) includes provision for renewal of all exploitation licences for a successive period of 15 years, providing the holder has not breached the licence obligations of license fee an annual surface rights fees payment and upholds all environmental standards set out in the exploitation Permit. Additionally, the Permit holder should provide the appropriate government departments with a monthly mining activity report and quarterly exploration reports.

The DRC Mining Code (2002) and Regulations have been amended with an updated Mining Code which came into force on 9th March 2018 (DRC Mining Code(2018)) and the related amended Mining Regulations which came into force on 8th June 2018.

Going forward the DRC Mining Code (2018) envisages a stability period for the tax, customs and exchange control regime of five years from the date on which the DRC Mining Code (2018) came into force and further provides that a number of the taxes shall be applied in accordance with the applicable substantive law.

The QP notes that the mining code is in the process of transition and the current proposed changes do not have an impact on the stated Mineral Resource or Ore Reserves at the gold prices used for Ore Reserve ($1,000/oz), Mineral Resource ($1,500/oz), or the current gold prices of $1,200/oz (August 2018).

All Permit fees, surface rights fees, and taxes relating to Kibali’s exploitation rights have been paid to date and reporting requirements conformed to, accordingly the concession is in good standing. At the time of compiling this report, the QP is not aware of any risks that could result in the loss of ownership of the deposits or loss of the permits, in part or in whole.

25.6 Infrastructure

Kibali is a mature operation that has all necessary support infrastructure already in place.

For purposes of reducing Kibali’s reliance on thermal generation and reducing the mine operating costs, three hydropower stations with a potential capacity of 44 MW of hydropower (at peak) and 32 MW of thermal Gen-sets. The average annual power consumption of the mine operation is approximately 40 MW, which the three hydropower stations are designed to supply approximately 80% of the mine demand taking into account fluctuation during the rainy seasons. Mine operating costs will be expected to be reduced in line with the introduction of hydropower station.

18^th September 2018 Page 335

Kibali Gold Mine NI 43-101 Technical Report

25.7 Risks

Kibali Goldmines has undertaken analysis of the Project risks. summarises the Project risks and the QPs assessment of the risk degrees and consequences, as well as ongoing/required mitigation measures. The QPs, however, note that the degree of risk refers to our subjective assessment as to how the identified risk could affect the achievement of the Project objectives.

Kibali has been in production for five years and is a mature operation

In the QP’s opinion, there are no significant risks and uncertainties that could reasonably be expected to affect the reliability or confidence in the exploration information, Mineral Resource or Ore Reserve estimates.

The following definitions have been employed by the QP s in assigning risk factors to the various aspects and components of the Project:

Risk Analysis Definitions

The following definitions have been employed by the QPs in assigning risk factors to the various aspects and components of the Project:

☐ Low – Risks that are considered to be average or typical for a deposit of this nature and could have a relatively insignificant impact on the economics. These generally can be mitigated by normal management processes combined with minor cost adjustments or schedule allowances.

☐ Minor – Risks that have a measurable impact on the quality of the estimate but not sufficient to have a significant impact on the economics. These generally can be mitigated by normal management processes combined with minor cost adjustments or schedule allowances.

☐ Moderate – Risks that are considered to be average or typical for a deposit of this nature but could have a more significant impact on the economics. These risks are generally recognisable and, through good planning and technical practices, can be minimised so that the impact on the deposit or its economics is manageable.

☐ Major – Risks that have a definite, significant, and measurable impact on the economics. This may include basic errors or substandard quality in the basis of estimate studies or project definition. These risks can be mitigated through further study and expenditure that may be significant. Included in this category may be environmental/social non-compliance, particularly in regard to Equator Principles and IFC Performance Standards.

☐ High – Risks that are largely uncontrollable, unpredictable, unusual, or are considered not to be typical for a deposit of a particular type. Good technical practices and quality planning are no guarantee of successful exploitation. These risks can have a major impact on the economics of the deposit including significant disruption of schedule, significant cost increases, and degradation of physical performance. These risks cannot likely be mitigated through further study or expenditure.

18^th September 2018 Page 336

Kibali Gold Mine NI 43-101 Technical Report

In addition to assigning risk factors, the QPs provided opinion on the probability of the risk occurring during the LOM. The following definitions have been employed by the QPs in assigning probability of the risk occurring:

☐ Rare – The risk is very unlikely to occur during the Project life.

☐ Unlikely – The risk is more likely not to occur than occur during the Project life.

☐ Possible – There is an increased probability that the risk will occur during the Project life.

☐ Likely – The risk is likely to occur during the Project life.

☐ Almost Certain – The risk is expected to occur during the Project life.

18^th September 2018 Page 337

Kibali Gold Mine NI 43-101 Technical Report

Risk Analysis Table

Table 25-1 details the Kibali Risk Analysis as determined by the QPs.

Table 25-1 Kibali Risk Analysis

Issue	Likelihood	Consequence Rating	Risk Rating	Mitigation
Geology and Mineral Resources – Confidence in Mineral Resource Models	Unlikely	Minor	Low	Additional scheduled infill drilling. Resource model updated on a regular basis using production reconciliation results.
Mining and Ore Reserves – Open Pit Slope Stability	Unlikely	Moderate	Minor	Continued in-pit monitoring, geotechnical drilling, instrumentation, and continued updating of geotechnical and hydrology models.
Mining and Ore Reserves – Underground Recovery and Dilution	Possible	Moderate	Low	Change in blasting practices to increase recovery and reduce dilution.
Processing				Several campaigns already successfully completed
- Water Dilution in CIL	Possible	Moderate	Medium	Plant modifications already installed to reduce water dilution at harvest screen and treatment of elution barrens
Processing Viscosity Drop in CIL Circuit Leading to carbon settlement	Possible	Moderate	Medium	Extensive trials performed. Viscosity modifiers or ensuring operation at acceptable slurry densities. Possible to increase the viscosity within the CIL by introducing oxides or flotation tails to supplement other ore or concentrate flows.
Environmental – Groundwater contamination (As) – Tailings failure	Possible	Major	Low	Manage As levels through feed profile. All high arsenic s feed reports to lined tailings facility. Continuing monitoring and external or third-party audits.
Social – Social License to Operate	Possible	Moderate	Moderate	Dedicated community engagement by company social and sustainability department.
Country & Political – Security – Governmental	Possible	Major	Moderate	Dedicated government liaison team in Kinshasa. Government participation/ownership.
Capital and Operating Costs	Unlikely	Moderate	Low	Continue to track actual costs and LOM forecast costs, including considerations for inflation and foreign exchange. Switching to Owner/Operator for underground mining in 2018
Fiscal Stability	Possible	Moderate	Moderate	Dedicated government liaison team in Kinshasa Government participation/ownership

18^th September 2018 Page 338

Kibali Gold Mine NI 43-101 Technical Report

26 Recommendations

The QPs make the following recommendations:

☐ A digital logging and data capture system to minimise manual data capture should be implemented.

☐ The 2018 mining sequence in the 5102 and lower 5101 zones has been modified. A full geotechnical review and model of mining induced stresses should be completed and if required the 2018 mining sequence adjusted.

☐ The following KPIs should be introduced for review at the time of the six-monthly update of the LOM plan to provide a quantifiable measure of the confidence in the change of plan:

o Number of continuous months in the LOM plan where 90% of scheduled stope production has had all resource grade control diamond drilling completed.

o Number of continuous months in the LOM plan where 90% of cross-cut and ore drive development has had all grade control diamond drilling completed.

☐ A professional development programme should be implemented aimed at developing suitably qualified resource geologists, mining engineers, and metallurgist to Qualified Person status.

☐ The option to extend the leach time in the main CIL circuit and improve process recoveries should be further considered.

☐ To further decrease the mine’s reliance on thermal power and potentially reduce operating costs, further sites suitable for hydropower generation in the Kibali region should be identified and subjected to independent feasibility studies.

☐ An ASM cessation strategy should be agreed with the Haut Uélé governor so that the local community and local chiefs are sensitised to the importance of limiting ASM activities within the government identified ‘corridors’.

☐ Active and direct engagement with the government regarding the impact of new proposed mining code

18^th September 2018 Page 339

Kibali Gold Mine NI 43-101 Technical Report

27 References

Allibone AH, Vargas C 2013: Southwest dipping fault at KCD pit. Unpublished memorandum to Randgold Resources Ltd, 6pp.

Allibone AH, Vargas C, Lawrence J. 2013: Geology and ore controls on cross section DGT040-DDD457 through the KCD Gold deposit, Kibali. Unpublished report to Randgold Resources Ltd, 45pp.

Allibone, AH. 2013. Controls on the location of gold mineralisation in the Kibali district, northeast DRC. Unpublished report to Randgold Resources Ltd, 39pp.

Allibone, AH. 2015: KZ Structure and mineralisation in the Kibali District. Unpublished report to Randgold Resources Ltd, 57pp.

Annual Closure Cost assessment, 2017

Beck Engineering (2014). Kibali Numerical Modelling Base Case Simulation. Letter Report dated

16 December 2014.

Beck Engineering (2015). Numerical Simulation of Kibali MHS. Report dated 10 September 2015.

Beck Engineering (2017). Global Deformation Modelling at Kibali. Report dated 22 January 2017.

Bird PJ, Treloar PJ, Vargas CA, Harbidge P, Millar I. 2014. The Kibali granite – greenstone belt: exploration and investigation of a new gold-bearing terrane. MDSG Poster.

Closure Liability 2017.

Coffey Mining (2013). Kibali Gold SA 3D Mine Wide Numerical Modelling – Stage 1. Report dated

11 March 2013.

Coffey Mining (2014). Kibali Gold SA 3D Mine Wide Numerical Modelling – Stage 2. Report dated

31 March 2014.

Community Development Plan (Undated)

Competent Persons Report Mineral Resources, Kibali Gold Mines, DRC. Compiled by Simon West, Project Resource Geologist, Randgold Resources Limited, 31 December 2017.

Competent Persons Report, Kibali Gold Mine 2017 Open Pit Ore Reserve Statement, December 2017.

CSR Strategy May 2017.

18^th September 2018 Page 340

Kibali Gold Mine NI 43-101 Technical Report

Cube Consulting Pty Ltd (2009), Amended and Restated Technical Report (Ni 43-101), Moto Gold Project Democratic Republic of Congo for Moto Goldmines Ltd. April 2009

Davis, B. 2004. Moto Project, Structural Geological Investigation. Unpublished Report to Moto Goldmines Limited, 40 pp.

December 2017 Dewatering Review (PowerPoint), Mark Raynor, SRK Consulting, December 2017.

Dempers & Seymour (2012). Kibali Project Mining Rock Mass Model. Report dated November 2012.

Dempers & Seymour (2014). Kibali Project Mining Rock Mass Model Update. Report dated November 2014.

Dempers & Seymour (2015). Kibali Project Mining Rock Mass Model Update. Report dated March 2015.

Dempers & Seymour (2017). Kibali Project Mining Rock Mass Model Update. Report dated March 2017.

Dempers & Seymour (2017). Kibali Project Mining Rock Mass Model Update. Report dated September 2017.

Environmental Incident Register 2017.

Global Deformation Modelling at Kibali, Beck Engineering, January 2017.

Gorumbwa RAP Progress Report May 2018 (ppt)

Gorumbwa RAP Report May 2018.

Grievance Mechanism Procedure.

Grievance Register 2017.

ISO 14001:2015 (EMS) Certificate, Feb 2018.

JORC Code 2012 Edition, Australasian Code for Reporting of Exploration Results, Mineral Resources and Ore Reserves, Joint Ore Reserve Committee of The Australian Institute of Mining and Metallurgy, Australian Institute of Geoscientists and Mineral Council of Australia (JORC), 2012.

Kibali ESIA 2011.

Kibali ESIA Update 2016.

18^th September 2018 Page 341

Kibali Gold Mine NI 43-101 Technical Report

Kibali RAP 2012.

Kibali Gold Mine 2016 Ore Reserve Report. Piran Mining Pty Ltd. Report dated February 2017.

Kibali Mineral Resource and Ore Reserve process review – August - December 2017, Optiro Pty Ltd, February 2018.

Kibali Mineral Resource Validation – December 2017, Optiro Pty Ltd, January 2018.

Kibali Mineral Resource Validation – December 2017. Optiro.

Kibali Optimised Feasibility Study, Randgold Resources Limited, June 2012.

Kibali Ore Reserve audit, Project no AU4312, Snowden Mining Industry Consultants, February 2014.

Kibali Project Mining Rock Mass Model update, Dempers and Seymour, March 2017.

Kibali Register of Permits, Licences and Authorisations.

Kibali Resettlement Audit March 2013.

Kibali Water Quality Data Review April 2016.

KSCA Geomechanics Pty Ltd (2012). A Review of the SRK Consulting Kibali Underground Geotechnical Feasibility Study Report (Rev0) dated February 2012.

KSCA Geomechanics Pty Ltd (2016). Kibali Stope Performance Database. Excel spread sheet dated 3 September 2016.

KSCA Geomechanics Pty Ltd (2017). Kibali Gold Mine Stope Performance (Stability Graphs). Report dated April 2017.

LOM Stakeholder Engagement Plan July 2015.

Mineral Resource Review, Kibali DRC, Quantitative Group Pty Ltd, Project code RRS21301, March 2013.

NI 43-101 Technical Report (dated 20th May 2010) by Adams et al., Kibali Goldmines SA

Nov 2011 Section of FS on Permitting and Environmental

NQA EMS Audit Report, Dec 2017.

Nzoro ESIA 2011.

Quarterly Report Q1 2018.

18^th September 2018 Page 342

Kibali Gold Mine NI 43-101 Technical Report

Quarterly Report Q4 2017.

Social Licence Strategy 2016.

SRK Consulting (2011). Kibali Underground Geotechnical Feasibility Study report (Rev0) dated November 2011.

Western Australian Department of Mines and Petroleum. Western Australian Mines Safety and Inspection Regulations 1995.

Western Australian School of Mines (2012). Stress Measurements from Oriented Core using the Acoustic Emission Method. Report dated December 2012.

18^th September 2018 Page 343

Kibali Gold Mine NI 43-101 Technical Report

28 Date and Signature Page

This report titled “Technical Report on the Kibali Gold Mine, Democratic Republic of the Congo”, effective 31 December 2017 and dated 18 September 2018, was prepared and signed by the following authors:

	(Signed & Sealed) “Rodney B. Quick”
Dated at London, UK	Rodney B. Quick, MSc, Pr. Sci.Nat,
18th September 2018	Group General Manager of Evaluations
	Randgold Resources Ltd.
	(Signed & Sealed) “Simon P. Bottoms”
Dated at London, UK	Simon P. Bottoms, CGeol, MGeol, FGS,
18th September 2018	MAusIMM
	Group Mineral Resource Manager
	Randgold Resources Ltd.
	(Signed & Sealed) “Richard Quarmby”
Dated at London, UK	Richard Quarmby, Pr Eng, C Eng,
18th September 2018	MSAIChE, MIoMM, MBA
	Group Metallurgist
	Randgold Resources Ltd.
	(Signed & Sealed) “Andrew Law”
Dated at Perth, Australia	Andrew Law, HND(MMin), MBA,
18th September 2018	FAusIMM (CP), FIQA, MAICD, AFAIM,
	Associate Director of Mining
	Optiro Pty Ltd
	(Signed & Sealed) “Graham E. Trusler”
Dated at Johannesburg, SA	Graham E. Trusler, MSc, Pr Eng,
18th September 2018	MIChE, MSAIChE
	Chief Executive Digby Wells and
	Associates Pty Ltd.

18^th September 2018 Page 344

Kibali Gold Mine NI 43-101 Technical Report

29 Certificate of Qualified Persons

29.1 Simon P. Bottoms, CGeol, MGeol, FGS, MAusIMM

I, Simon P. Bottoms, CGeol, MGeol, FGS, MAusIMM as, an author of this report entitled “Technical Report on the Kibali Gold Mine, Democratic Republic of the Congo”, effective 31^st December 2017 and dated 18^th September 2018, do hereby certify that:

1. I am the Group Mineral Resource Manager and an Officer of Randgold Resources Limited 3rd Floor, Unity Chambers, 28 Halkett Street, St. Helier, Jersey, Channel Islands.

2. I graduated with a Masters of Geology degree from the University of Southampton, United Kingdom in 2009.

3. I am a Chartered Geologist registered (1023769) with the Geological Society of London. I am also a current Member of AusIMM. I have worked as a geologist continuously for 9 years since my graduation from University.

4. I have read the definition of “Qualified Person” set out in National Instrument 43-101 (NI 43-101) and certify that by reason of my education, affiliation with a professional association (as defined in NI 43-101) and past relevant work experience, I fulfil the requirements to be a “Qualified Person” for the purposes of NI 43-101.

5. I have been involved at Kibali Gold Mine since 2012 and regularly visit the site. I most recently visited the Kibali Gold Mine from 16th to 27th August 2018.

6. I am responsible for the preparation of sections 4 to 12, 14, and 23 of this Technical Report. I share responsibility with my co-authors for sections 1, 2, 3, 24, 25, 26, and 27.

7. I am not independent of the Issuer applying the test set out in Section 1.5 of NI 43-101 since I am a full time employee at Randgold Resources Limited.

8. I have had prior involvement with the property that is the subject of the Technical Report. I am a full time employee at Randgold Resources Limited and I have been involved with Kibali since 2012.

9. I have read NI 43-101 and Form 43-101F1 and the Technical Report has been prepared in compliance with that instrument and form.

10. As of the date of this certificate, to the best of my knowledge, information and belief, the sections 1 to 4, 6 to 12, 14 and 23 to 27 of this Technical Report contains all scientific and technical information that is required to be disclosed to make the Technical Report not misleading.

Dated this 18^th day of September 2018.

(Signed & Sealed) “Simon P. Bottoms”

Simon P. Bottoms, CGeol, MGeol, FGS, MAusIMM

18^th September 2018 Page 345

Kibali Gold Mine NI 43-101 Technical Report

29.2 Rodney B. Quick, MSc, Pr. Sci.Nat

I, Rodney B. Quick, MSc, Pr. Sci.Nat, as an author of this report entitled “Technical Report on the Kibali Gold Mine, Democratic Republic of the Congo”, effective 31^st December 2017 and dated 18^th September 2018, do hereby certify that:

I am the Group General Manager of Evaluations and an Officer of Randgold Resources Limited 3rd Floor, Unity Chambers, 28 Halkett Street, St. Helier, Jersey, Channel Islands.

1. I graduated with a Bachelor of Science Honours degree in Geology from the University of Natal Durban, South Africa in 1993, and with a Master of Science degree in Geology from the Leicester University, United Kingdom in 2000.

2. I am a Professional Natural Scientist registered (400014/05) with the South African Council for Natural Scientific Professions (SACNASP). I am a current Member of SACNASP. I have worked as a geologist for 24 years since my graduation from University.

3. I have read the definition of “Qualified Person” set out in National Instrument 43-101 (NI 43-101) and certify that by reason of my education, affiliation with a professional association (as defined in NI 43-101) and past relevant work experience, I fulfil the requirements to be a “Qualified Person” for the purposes of NI 43-101.

4. I have been involved at Kibali Gold Mine since 2009 and regularly visit the site. I most recently visited the Kibali Gold Mine from 10^th to 14^th September, 2018.

5. I am responsible for the preparation of sections 19, and 22 of this Technical Report. I share responsibility with my co-authors for sections 1, 2, 3, 21, 24, 25, 26, and 27.

6. I am not independent of the Issuer applying the test set out in Section 1.5 of NI 43-101 since I am a full time employee at Randgold Resources Limited.

7. I have had prior involvement with the property that is the subject of the Technical Report. I am a full time employee at Randgold Resources Limited and I have been involved with Kibali since 2009.

8. I have read NI 43-101 and Form 43-101F1 and the Technical Report has been prepared in compliance with that instrument and form.

9. As of the date of this certificate, to the best of my knowledge, information and belief, the sections 1, 2, 3, 19, 21, 22, 24, 25, 26 and 27 of this Technical Report contains all scientific and technical information that is required to be disclosed to make the Technical Report not misleading.

Dated this 18^th day of September 2018.

(Signed & Sealed) “Rodney B. Quick”

Rodney B. Quick, MSc, Pr. Sci.Nat

18^th September 2018 Page 346

Kibali Gold Mine NI 43-101 Technical Report

29.3 Richard Quarmby, BSc, Pr Eng, C Eng, MSAIChE, MIoMMM, MBA

I, Richard Quarmby, BSc, Pr Eng, C Eng, MSAIChE, MIoMMM, MBA, as an author of this report entitled “Technical Report on the Kibali Gold Mine, Democratic Republic of the Congo”, effective 31^st December 2017 and dated 18^th September 2018, do hereby certify that:

1. I am the Group Metallurgist - Projects and an Officer of Randgold Resources Limited 3rd Floor, Unity Chambers, 28 Halkett Street, St. Helier, Jersey, Channel Islands.

2. I graduated with a BSc chemical engineering degree from the University of the Witwatersrand in 1985 and earned a Master of Business Administration degree in 2005.

3. I have been registered, no. 910237 as a Professional Engineer (Pr Eng) with the Engineering Council of South Africa since 1991 and in 2010 was accepted to the UK equivalent institution i.e. Chartered Engineer with the Engineering Council UK (C Eng), no. 580441. Further, I have been a Member, no. 1361, of the South African Institution of Chemical Engineers (SAIChE) since 1989, and am also a registered Member, no. 454225, of the Institute of Materials, Minerals and Mining (IoMMM) UK. I have worked as an engineer continuously since my graduation from University in 1985.

4. I have read the definition of “Qualified Person” set out in National Instrument 43-101 (NI 43-101) and certify that by reason of my education, affiliation with a professional association (as defined in NI 43-101) and past relevant work experience, I fulfil the requirements to be a “Qualified Person” for the purposes of NI 43-101.

5. I have been involved at Kibali Gold Mine since 2016 and regularly visit the site. I most recently visited the Kibali Gold Mine from January to April, 2017.

6. I am responsible for the preparation of sections 13, 17 and 18 of this Technical Report. I share responsibility with my co-authors for sections 1, 2, 3, 21, 25, 26, and 27.

7. I am not independent of the Issuer applying the test set out in Section 1.5 of NI 43-101 since I am a full time employee at Randgold Resources Limited.

8. I have had prior involvement with the property that is the subject of the Technical Report. I am a full time employee at Randgold Resources Limited and I have been involved with Kibali since 2016.

9. I have read NI 43-101 and Form 43-101F1 and the Technical Report has been prepared in compliance with that instrument and form.

10. As of the date of this certificate, to the best of my knowledge, information and belief, the sections 13, 17 and 18 of this Technical Report contains all scientific and technical information that is required to be disclosed to make the Technical Report not misleading.

Dated this 18^th day of September 2018.

(Signed & Sealed) “Richard Quarmby”

Richard Quarmby, BSc, Pr Eng, C Eng, MSAIChE, MIoMMM, MBA

18^th September 2018 Page 347

Kibali Gold Mine NI 43-101 Technical Report

29.4 Andrew Law, HND (MMin), MBA, FAusIMM (CP), FIQA, MAICD, AFAIM

I, Andrew Law, HND(MMin), MBA, FAusIMM (CP), FIQA, MAICD, AFAIM, as an author of this report entitled “Technical Report on the Kibali Gold Mine, Democratic Republic of the Congo”, effective 31^st December 2017 and dated 18^th September 2018, do hereby certify that:

1. I am Associate Director of Mining for Optiro Pty Ltd, of Perth, Australia.

2. I graduated in 1983 with a Higher National Diploma in Mine Engineering (Witwatersrand), and in 2007 with a Master of Business Administration degree (University of Western Australia).

3. I am a Fellow of the AusIMM (CP) (no: 107318), a Fellow of the Institute of Quarrying – Australia, a Member of the Australian Institute of Company Directors and an Associate Fellow of the Australian Institute of Management. I have worked as a mining engineer continuously since my graduation from University in 1983.

4. I have read the definition of “Qualified Person” set out in National Instrument 43-101 (NI 43-101) and certify that by reason of my education, affiliation with a professional association (as defined in NI 43-101) and past relevant work experience, I fulfil the requirements to be a “Qualified Person” for the purposes of NI 43-101. I have over 34 years of experience within the mining industry in technical, management, and consulting positions.

5. I have been involved at Kibali Gold Mine since 2017, I most recently visited the Kibali Gold Mine in July 2017 and in December 2017.

6. I am responsible for the preparation of sections 15, and 16 of this Technical Report. I share responsibility with my co-authors for sections 1, 2, 3, 25, 26, and 27.

7. I am independent of Randgold Resources Ltd. and related companies applying the test set out in Section 1.5 of NI 43-101.

8. I have not had prior involvement with the property that is the subject of the Technical Report.

9. I have read NI 43-101 and Form 43-101F1 and the Technical Report has been prepared in compliance with that instrument and form.

10. As of the effective date of this report, to the best of my knowledge, information and belief, sections 15, and 16, and parts of sections 1, 2, 3, 25, 26, and 27 of this Technical Report for which I am responsible contains all scientific and technical information that is required to be disclosed to make this Technical Report not misleading.

Dated this 18^th day of September 2018.

(Signed & Sealed) “Andrew Law”

Andrew Law, HND(MMin), MBA, FAusIMM (CP), FIQA, MAICD, AFAIM

18^th September 2018 Page 348

Kibali Gold Mine NI 43-101 Technical Report

29.5 Graham E. Trusler, MSc Pr Eng, MIChE, MSAIChE

I, Graham E. Trusler, MSc, Pr Eng, MIChE, MSAIChE as an author of this report entitled “Technical Report on the Kibali Gold Mine, Democratic Republic of the Congo”, effective 31 December 2017 and dated 18 September 2018, do hereby certify that:

1. I am the CEO of Digby Wells and Associates Pty Ltd., of Johannesburg, South Africa.

2. I graduated with a Masters of Chemical Engineering degree from the University of KwaZulu-Natal, South Africa, in 1988.

3. I have been registered, no. 920088 as a Professional Engineer (Pr Eng) with the Engineering Council of South Africa since 1992. Further, I have been a Member, of the South African Institution of Chemical Engineers (SAIChE) since 1994. I am also registered as a Chartered Chemical Engineer with the Institution of Chemical Engineers, is a member of the Water Institute of South Africa and a lifetime member of the American Society of Mining and Reclamation. I have worked as an engineer continuously from 1990.

4. I have read the definition of “Qualified Person” set out in National Instrument 43-101 (NI 43-101) and certify that by reason of my education, affiliation with a professional association (as defined in NI 43-101) and past relevant work experience, I fulfil the requirements to be a “Qualified Person” for the purposes of NI 43-101. I have over 30 years of experience within the mining industry in metallurgical production, research and environmental issues.

5. I have been involved at Kibali Gold Mine since 2010 and have visited the site on numerous occasions. I most recently visited the Kibali Gold Mine from 15 to 18 May, 2017.

6. I am responsible for the preparation of section 20 of this Technical Report. I share responsibility with my co-authors for sections 1, 2, 3, 25, 26, and 27.

7. I am independent of the Randgold Resources Ltd. and related companies applying the test set out in Section 1.5 of NI 43-101.

8. I have had prior involvement with the property that is the subject of the Technical Report.

9. I have read NI 43-101 and Form 43-101F1 and the Technical Report has been prepared in compliance with that instrument and form.

10. As of the effective date of this report, to the best of my knowledge, information and belief, sections 20, and parts of sections 1, 2, 3, 25, 26, and 27 of this Technical Report for which I am responsible contains all scientific and technical information that is required to be disclosed to make this Technical Report not misleading.

Dated this 18^th day of September 2018.

(Signed & Sealed) “Graham E. Trusler”

Graham E. Trusler, MSc, Pr Eng, MIChE, MSAIChE

18^th September 2018 Page 349

Kibali Gold Mine NI 43-101 Technical Report

30  Appendix

30.1 Appendix 1 – JORC 2012 Edition – Table 1

The following table provides a summary of important assessment and reporting criteria used at the Kibali Mine for the reporting of Mineral Resources and Ore Reserves in accordance with the Table 1 checklist in The Australasian Code for the Reporting of Exploration Results, Mineral Resources and Ore Reserves (The JORC Code, 2012 Edition). Criteria in each section apply to all preceding and succeeding sections.

Section 1. Sampling Techniques and Data

JORC (2012) Code Checklist of Assessment and Reporting Criteria

Criteria	Commentary
Sampling techniques	☐	Diamond drill (DD) half core samples are taken within geological units, and are normally between 0.8 and 1.2m long. The diamond drilling has been completed by Boart Longyear (surface DD) and Ore Zone (underground DD), dominantly in NQ size. Diamond core is photographed before being halved with a diamond saw. One half is submitted for assay analysis whilst the other half is stored for future reference.
	☐	Reverse circulation (RC) samples are riffle split and composited on 2m downhole samples. RC drilling is completed by Boart Longyear and Ore Zone.
	☐	Rotary air blast drilling (RAB) drilling is used in regional first pass exploration and for sterilisation purposes. No samples used for resource estimation are from RAB holes.
	☐	Chip samples are used within the underground development area to provide an additional source of information regarding the mineralisation associated with the alteration, particularly when mapping low-grade halo contacts. This data is recorded on the underground geological maps, which are then scanned and georeferenced for wireframe model updating. However, this data is not used for estimation.
	☐	Others sample such as grab sample, channel samples and soil samples are also used based on the area of interest and stage of exploration.
Drilling techniques	☐	DD is utilised for resource extension work. PQ rods (85.0 mm) are generally used for the first 100 m down hole with HQ (63.3 mm) or NQ (47.6 mm) used from 100 m to 200 m with depending on the drilling depth requirement. All underground grade control and Advanced Grade Control; diamond drilling is completed in NQ.
	☐	RC holes are used for advanced grade control and infill grade control drilling using 131 mm diameter rods. RC chips are sieved and logged

18^th September 2018 Page 350

Kibali Gold Mine NI 43-101 Technical Report

Criteria	Commentary
		by the geologist before being placed in chip boxes for storage.
	☐	All drill holes have had their collar location surveyed using differential GPS to a 10 mm accuracy.
	☐	Prior to mid-2016 Reflex EZ-Trac tools were used but were replaced by Reflex EZ-Gyro. All drill holes are now surveyed down hole with a Reflex EZ-Gyro.
Drill sample recovery	☐	Core recovery is measured in the field and during detailed logging; core loss is marked out clearly.
	☐	RC sample recovery is measured by weighing the total weight of sample collected over a meter drilled and comparing it to the theoretical expected weight for each material type (lithological unit) and weathering type.
	☐	Sample size optimisation has been completed for each deposit and the KCD deposit uses 2 m RC samples, this is reviewed on each deposit to verify the suitability of the sample size.
Logging	☐	DD core is geologically logged and includes weathering, mineralisation, alteration, lithology, structure and redox. This is stored in a central database after validation.
	☐	All diamond drill core is oriented and where orientation is not possible the core is assembled with previous runs where possible to try and extend the orientation line, such that structural directions in the form of alpha and beta angles are documented.
	☐	Geotechnical logging is only performed for holes drilled specifically for geotechnical assessment as required.
	☐	RC chips samples are logged with the same lithological, mineralogical and alteration information as DD core but are logged on the 2 m RC samples from the riffle splitter.
	☐	Logging is completed on hard copy before being transcribed to the database.
Sub-sampling techniques and sample preparation	☐	Half core is submitted for assay wherever possible, quarter core is only used when there is a need for further assays or other analysis to be completed which need to be completed on the same interval.
	☐	RC samples are collected from the rig in two metre intervals using a riffle splitter. Auxiliary booster units are used to ensure the vast majority of the samples collected are already dry. On the rare occasion a wet sample is obtained it is dried before being manually split.
Quality of assay data and laboratory tests	☐	All samples for the Project are analysed by SGS Doko laboratory at the Kibali mine site or at the SGS Mwanza laboratory in Tanzania, which was used as an overflow laboratory, multi element analysis and soils analysis.
	☐	Both laboratories are certified and operated independently by SGS. All samples are analysed using lead collection 50 g fire assay with atomic absorption (AA) finish and a gravimetric (GA) finish for any samples reporting above 100 g/t Au.
	☐	Quality control checks are inserted into sample stream prior to dispatch to the laboratory except for diamond core duplicates which are taken as a split by Kibali staff in the laboratory using a riffle splitter after crushing but before pulverising. Overall, the QA/QC sampling includes 5% field duplicates, 5% blanks and 5 % Certified Referenced Materials (CRM).

18^th September 2018 Page 351

Kibali Gold Mine NI 43-101 Technical Report

Criteria	Commentary
	☐	Secondary labs are used on quarterly basis as check for the primary laboratory as well as to check the consistency in sampling protocols.
	☐	During 2017: 249,359 samples were submitted for analysis at either SGS Doko or SGS Mwanza. 12,497 of these were CRMs, 12,140 were blanks and 12,423 were field duplicates
	☐	Coarse blanks samples are composed of barren granite material from 20 km NW of the Project area and prepared on site. In 2018 blank sample raw material will be replaced with OREAS certified blanks that reflect the geology on site.
	☐	Field duplicates analysed at SGS Doko and SGS Mwanza show a good correlation. Some issues are observed in very high-grade samples which are likely a result of the nugget effect. The overall impact of this variation is negligible due to the low population at these grades.
Verification of sampling and assaying	☐	Twin holes are used to access the accuracy of some intersections that require further details.
	☐	All information and data captured are stored in central database (Maxwell DataShed) model with validation tool and regular checks in place.
	☐	All forms of Project data are stored secured in industry standard Maxwell DataShed SQL database for optimal validation through constraints, library tables, triggers, and stored procedures. Any data that fails validation is rejected and stored in a buffer table awaiting correction.
	☐	A custom MS Access front end application has been designed for data entry, reporting and viewing via ODBC, which utilises the data validation procedures from the SQL database.
	☐	Any site software application databases link back to the main database for information retrieval via ODBC.
	☐	Assay data is imported directly from assay certificates from the laboratory and validated. Assay data is stored in a normalised format and multiple assays are stored for each sample. Ranking of different assay formats is performed automatically so that one assay result is displayed in the tblVWDHAssays table. Any change to the rankings (held within tblSYSAssMethod) must be approved by the onsite Database Manager.
Location of data points	☐	All collared information is measured with Trimble differential GPS with an accuracy of 10 mm (working Grid WGS84 Zone 35N)
	☐	Down hole survey is carried out using both Reflex EZ-GYRO, Reflex EZ-Trac and conventional Gyro instruments. The EZ-Trac survey tools were phased out during 2016 for replacement with the EZ-GYRO tools – which are not susceptible to any magnetic interference. Previously when both EZ-Trac and conventional Gyro surveys were being completed, the Gyro survey takes a higher priority over Reflex EZ-TRAC surveys.
	☐	The topographic surface was last updated in December 2010. A 2 m contoured Lidar surface is used which has been validated and high level of accuracy found with known control points.
	☐	Active mining areas are scanned using laser scanners on a monthly basis and detailed drone photometry surface scans are completed on a weekly basis.
	☐	Down hole survey equipment is calibrated yearly, and checked every quarter by Reflex technicians during site visits.

18^th September 2018 Page 352

Kibali Gold Mine NI 43-101 Technical Report

Criteria	Commentary
Data spacing and distribution	☐	RC open pit grade control drill holes (Measured Resources) have independently optimised drill spacing using change of support analysis. However, in general the infill drill spacing ranges between 10 m and 20 m along the principal direction and 5-10 m across strike within the ore zones and are sampled at 2 m downhole intervals.
	☐	Resource drill holes (indicated) also have independently optimised drill spacings using change of support analysis. However, in general they are spaced approximately 40 m by 40 m with geological continuity of 100 m and more along strike. All open pit resources that also form reserves, namely KCD; Kombokolo; Pakaka; Pamao; Gorumbwa and Sessenge have been drilled to an advanced grade control spacing.
	☐	Inferred Resource drill holes are on an 80 m by 80 m or less drill spacing.
	☐	The data distribution is one of several classification specifications for the resource estimate which includes minimum sample points, kriging variance and conditional bias.
	☐	All drilled holes are composited to 2 m down hole during resource estimation; this is supported by sample interval optimisation study completed as part of the feasibility that demonstrated 2 m is optimal for sampling within all known deposits in the Kibali Permit at that time.
Orientation of data in relation to geological structure	☐	Drilling directions for updated Mineral Resources namely Pakaka, Kombokolo, Sessenge, Pamao, Gorumbwa, Mengu Hill, and KCD open pit are optimised on an individual deposit basis to ensure that the preferred drilling direction for Resource and Grade Control, drilling is on a cross plunge basis.
	☐	KCD underground Mineral Resource and geological model has been significantly affected by a sampling bias with the feasibility drilling data being conducted from surface. This has meant that some of the sub vertical lodes such as 9105 and 5101 were initially delineated using sub optimal drill directions. Since 2016 there has been a significant quantity of advanced GC drilling conducted from the underground development – where the drilling could be completed with perpendicular angles of intersection to the primary 9105, 5101 & 5110 ore lodes. The results of this drilling have significantly improved the modelled definition of the banded ironstone as the marker unit for the km-scale NE plunging fold structure, which acts as the primary control on the positioning of the 5000 and 9000 ore lodes and delineated zones of internal waste within the 9105. This has resulted in a significant model change affecting both the 2017 Mineral Resources and Ore Reserves.
	☐	Significant additional Advanced Grade control drilling will continue to be conducted throughout 2018, it is anticipated that this drilling will define further model changes in areas not yet tested, particularly in the lower portion and down plunge of the 9105 lode, although additional work since the 2016 resource model has confirmed the updated model specifically in the upper portions of the 5101 & 5110 lodes. However, further model changes have been seen in the down plunge portions of the 5101 which have not previously included any advanced grade control drilling. As at the Mineral Resource cut-off all material model changes are incorporated; however further changes were reported as of EOM November GC model. These changes will be incorporated into the 2018 Mineral Resource and have already been factored in within the 2017 Ore Reserves.
	☐	The 2016 resource report highlighted the risk to the 9101 resource due to the lack of additional drilling since completion of the feasibility, but the angle of intersection was favourable being near perpendicular. During 2017, advanced grade control drilling was completed and confirmed the mineralisation of 9101 and interpreted to join the 9105, based on the folded BIF model. Further infill grade control drilling is planned for 2018 to more accurately define the local grade variability prior to commencement of mining the 9101 in 2019.

18^th September 2018 Page 353

Kibali Gold Mine NI 43-101 Technical Report

Criteria	Commentary
Sample security	☐	Pulp samples are stored under condition that are kept clean and dry to avoid contamination.
	☐	Samples are under security observation from collection at rig, to processing at the site core yard, to delivery at the laboratory.
	☐	Samples are weighed and documented on the rig.
Audits or reviews	☐	Optiro completed an audit of the Mineral Resource process in 2017, which found no essential issues to be fixed. A number of recommended issues were identified, but have all been addressed by September 2017.
	☐	The RC subsampling practice needed to be revised to ensure consistent sample weights within the tolerance (3 kg to 3.5 kg) required by the SGS Doko assay laboratory. A new set of Gilson splitters were procured and installed onsite, thereby ensuring an unbiased and consistent subsample.
	☐	In line with the above, once RC samples of the right mass are being delivered to the laboratory they should all be processed through the Boyd crusher and associated rotary sample divider. The laboratory should not need to split the samples using a riffle splitter as is current practice.
	☐	Blank samples (granite fragments) should be selectively inserted within the high-grade intersections as logged, whether for diamond core or RC chips. This will ensure maximum effectiveness and value for money of the blank insertion process.

18^th September 2018 Page 354

Kibali Gold Mine NI 43-101 Technical Report

Section 2. Reporting of Exploration Results

(Criteria listed in the preceding section also apply to this section.)

Criteria
Mineral tenement and land tenure status	☐	Kibali Goldmines has been granted ten Exploitation Permits under the DRC Mining Code (2002) in respect of the Kibali Gold Project, eight of which are valid until 2029 and two of which are valid until 2030.
	☐	Kibali Goldmines is joint venture Randgold Resources and AngloGold Ashanti with each controlling a 45% stake. The remaining 10% is owned by Congolese parastatal Société des Mines d’Or de Kilo-Moto (SOKIMO).
Exploration done by other parties	☐	Gold was first discovered in the territory in 1903. The Kibali Project was originally discovered by a JV between Barrick and AngloGold in 1998. Barrick completed limited drilling of the Project during this period. In 2004 Moto Goldmines acquired the Kibali Project and discovered the main KCD deposit. Moto Goldmines completed a significant amount of DD, RC as well pitting and trenching. Randgold Resources and AngloGold Ashanti acquired Moto Goldmines in 2009.
	☐	All historical drilling completed by Barrick or Moto is reviewed on a case by case basis, prior to undertaking any key geological model reviews or updates of a Mineral Resource. Initially a number of old holes are selected for twinning and this data is then used to make an informed decision as to the reliability of the old data on that Mineral Resource. In general the twins completed to date have shown that assayed intercepts are mostly repeatable. However, some twins have identified that the ore intercept is at a different depth down hole relative to the historic data, thereby indicating that either the down hole survey or collar survey data of the historic data is not reliable. In any instances where this has been observed the historical holes have been removed from updated estimation datasets. However, there are still a number of Inferred satellite resources which rely heavily on the historic data namely Pamao, Megi, Marakeke & Mengu Village, which currently poses a moderate level of risk to these Mineral Resources. Historical maps have also been used in the definition of old underground mine workings particularly at Gorumbwa and Durba Hill. The Gorumbwa historical mine workings have been broadly confirmed with a phase of resource definition drilling during 2016 and 2017, which reduces the risk of a significant impact on the economic viability of the Mineral Resource. However, the detailed definition of this void still poses a potential safety risk to the operation and more detailed void survey is planned for 2018. At Durba hill, survey scanning and probe drilling was completed in 2017.
Geology	☐	The deposits differ from many orogenic gold deposits in terms of structural setting. Rather than being linked to a major large scale steeply dipping strike slip fault with brittle-ductile deformational evolution, they are hosted within a thrust stack sequence with ductile to brittle- ductile deformational structures and complex folding history
	☐	Most gold mineralisation hosted by the KCD and satellite deposits is texturally associated with fine disseminated pyrite, with minor pyrrhotite and arsenopyrite. The auriferous pyrite occurs as both ‘salt and pepper’ disseminated fine grains and clusters of disseminated grains forming blebs and pseudo-vein mosaics. Petrographic study has identified several sulphide phases with arsenopyrite, chalcopyrite, pyrrhotite and pyrite dominating the assemblage with multiple generations of each identified. Gold is hosted within the dominant second

18^th September 2018 Page 355

Kibali Gold Mine NI 43-101 Technical Report

Criteria
		pyrite phase and as late fracture fillings associated with chalcopyrite and galena. The gold bearing pyrite is hosted by the sequence of coarser clastic sedimentary unit’s conglomerate and chert-ironstone assemblage, often with an envelope of ACSA alteration.
	☐	The Gorumbwa lode plunges at a low to moderate angle to the NE. The deposit was mined by SOKIMO commencing in 1955 from underground and small open pit operations. Total production was estimated at approximately 2.8 Mt at approximately 7 g/t Au. The underground and open pit workings are presently collapsed and flooded. Dewatering of the historic open pit is in progress to facilitate drilling programs. Two historic vertical shaft head frames remain to the east of the historic pit. Underground workings extend to 380 m below surface. The mineralisation consists of a series of stacked ‘lenses’ that variably extend down plunge for a length of 1,000 m at an average width of 200 m and have been mined to a depth of 400 m below topographic surface.
	☐	Gold mineralisation at Pakaka-Pamao is hosted by the met conglomerate interbedded with minor tuffaceous units. Recent drilling and trenches completed in 2017 shows mineralisation to be hosted in ironstone and sandstones in Pamao. The mineralised zones are characterised by silica-ankerite-pyrite alteration, mainly in well foliated siliceous rocks. The mineralised zones are associated with pervasive silicification with local preservation of breccia textures that have been overprinted by the dominant S1 fabric. Higher gold grades appear to correlate well with the presence and abundance of pyrite, and empirically the high-grades appear to be spatially associated with the intersection of the NW trending D1 thrust surface, and a NE trending strain corridor. The structures combine to produce a broad NE plunging open anticlinal structure, with Pamao on the west limb, and Pakaka on the east. The Pakaka mineralisation continues down plunge beyond the limits of the drilling and represents a further exploration potential. The Pakaka mineralisation extends over a strike length of 1,000 m, averages a thickness of 30 m and has been identified to a depth of 350 m below surface. The weathering profile at Pakaka is relatively deep.
Drill hole Information	☐	No exploration results are reported in this document.
Data aggregation methods	☐	Individual exploration results are seldom reported for Kibali mine. Where applicable any results are capped to the same grades as those used for the relevant geological domain defined within the existing Mineral Resource.
Relationship between mineralisation widths and intercept lengths	☐	No exploration results are reported in this document.
	☐	Drilling on Kibali Permit is orientated to ensure that the drilling intersections are as close to perpendicular as technically possible.
Diagrams	☐	No exploration results are reported in this document.

18^th September 2018 Page 356

Kibali Gold Mine NI 43-101 Technical Report

Criteria
Balanced reporting	☐	No exploration results are reported in this document.
Other substantive exploration data	☐	No exploration results are reported in this document.
Further work	☐	2018 resource definition work is scheduled to target extension of KCD underground resource/reserves and to continue the advanced grade control programme on a 60 m by 40 m definition. This drilling will be completed from underground to ensure that it is drilled perpendicular to the mineralisation so true thicknesses can be determined and internal waste inclusions identified. With planned exploration drive to be completed in 2018 to test the down plunge extension.
	☐	2018 KCD exploration will continue to target the up plunge extension of the 3000 lode, 9000 up plunge to Sessenge gap and the 5000 down plunge
	☐	Inferred satellite deposits and gaps between existing Mineral Resources are due to be ranked and re-evaluated for further testwork to upgrade the resources from conceptual or Inferred to a higher resource category and potentially definition of modifying factors for appropriate conversion to reserve.
	☐	KCD open pit resource for both 5000 and 3000 will also be further drill tested to upgrade the resource definition of the current Inferred resource and potentially further extend the existing resources and reserves.

18^th September 2018 Page 357

Kibali Gold Mine NI 43-101 Technical Report

Section 3. Estimation and Reporting of Mineral Resources

(Criteria listed in section 1, and where relevant in section 2, also apply to this section.)

Criteria	Commentary
Database integrity	☐	Data stored in (audited) Maxwell DataShed SQL database. Data must pass validation through constraints, library tables, triggers and stored procedures prior to importing. Failed data is either rejected or stored in buffer tables awaiting correction. Assay data is imported directly from laboratory certificates and only fully trained and authorised network users can upload laboratory data. All other MS Access databases on site use ODBC link to retrieve information from DataShed SQL database. A full-time database administrator employed at site manages the database.
	☐	Data loggers for collection of grade control lithological, logging and sampling data planned to be implemented.
Site visits	☐	Resource estimation is overseen by Mr Simon Bottoms, CGeol, MGeol, FGS, , MAusIMM, Group Mineral Resource Manager and a competent person and Mr Rodney Quick, MSc, Pr. Sci.Nat, SACNASP, General Manager Evaluation.
	☐	Mr Bottoms and Mr Quick visit site regularly to review exploration programme and results, Mineral Resource and grade control model updates, mine strategy and external auditing in addition to board reviews.
Geological Interpretation	☐	Geological paper cross sections are generated and georeferenced in software to be used as a basis for 3D modelling.
	☐	Geological interpretations are digitised as polylines on cross sections spaced 10 m apart on KCD. Geology, alteration, and low and high- grade polylines are snapped on each section to the corresponding sample interval. Polylines are wireframed between sections to build a valid 3D solid.
	☐	Mineralisation domains are sub domained into low-grade and high-grade, with domain analysis completed.
	☐	All domains use hard boundaries to ensure that separate grade populations do not influence the grades
	☐	The geological models are updated monthly, where grade control data is available, using input from the field geologists.
	☐	Interpretations are regularly cross checked with DD core and RC chips to ensure the model is representative.

18^th September 2018 Page 358

Kibali Gold Mine NI 43-101 Technical Report

Criteria	Commentary
		Deposit	Down Plunge (m)	Average       Width (m)	Plunge             Direction
Dimensions		KCD 3000	2,000	300	NE
		KCD 5000	1,500	100	NE
		KCD 9000	1,500	150	NE
		Mengu Hill	700	150	NNE
		Sessenge	500	250	NE
		Gorumbwa	800	70	NE
		Kombokolo	500	100	ENE
		Pakaka	1,000	350	NE
		Pamoa	900	450	NE
		Rhino	150	80	NE
		Megi	500	150	NE
		Marakeke	1,000	150	NW
		Mengu village	800	200	NW
Estimation and modelling techniques	☐	KCD-Sessenge, Mengu, Pakaka, Gorumbwa, Pamao, Megi and Kombokolo use Ordinary Kriging (OK) for estimation.
	☐	Marakeke, Mengu Village were estimated using Uniform Conditioning with an allowance for an Information Effect incorporating important modifying factors such as grade control drilling, mining selectivity and cut-off grade criteria.
	☐	Models are reconciled against the point data and against the previous models. SWATH plots, volume reconciliation, grade checks and visual validation techniques are used as check on the models.
	☐	Quantitative Kriging Neighbourhood Analysis (QKNA) was undertaken on each deposit to ensure suitable block sizes, search radius, discretisation and estimation parameters were used for the interpolation and thus minimise conditional bias. Block sizes used varies by drill spacing for different deposits.
	☐	Selective mining units are optimised for each individual Mineral Resource, which predominantly utilises the geological knowledge of the deposit to optimise the anticipated ore loss and dilution. However specific detailed studies are completed on areas of Mineral Resources with historic depletion voids, namely Gorumbwa & KCD.
	☐	At Pakaka a small amount of silver is within the ore that is submitted to the smelters for which a penalty is applied by the smelters.
	☐	Arsenic is estimated at Pakaka and Sessenge for geo-metallurgical purposes in order to optimise the process plant feed blend strategy.
	☐	Drill samples are composited down hole on a 2 m length between mineralisation boundaries. Residual samples below 0.5 m are excluded from the database used for resource interpolation.
	☐	Top cutting is applied to composited samples, using disintegration analysis, histogram and log probability plot. Top cutting values are reviewed for each mineralised domain individually and are generally between the 95th to 99.9th percentiles. Metal at risk simulations are planned to be reviewed during 2018.
Moisture	☐	Mineral Resource tonnages are estimated on a dry basis.

18^th September 2018 Page 359

Kibali Gold Mine NI 43-101 Technical Report

Criteria	Commentary
Cut off Parameters	☐	All open pit resources have been constrained within a $1,500/oz resource pit.
	☐	All Mineral Resources are reported at $1,500/oz insitu marginal cut-off grade analysis for that particular deposit. Each deposit has slightly different costs. Satellite pits generally will incur a haulage cost depending on the distance to the Rom Pad.
	☐	2016 KCD underground resources have been constrained at KCD underground using a mineable shape optimiser (MSO). Underground Mineral Resources for 2017 are insitu Mineral Resources that meet a cut-off of 1.6 g/t Au within a minimum mineable stope shape, reported at and a gold price of $1,500/oz.
Mining Factors or Assumptions	☐	Many of the deposits at Kibali are active mines and thus the economic and mining factors that could be used to determine a reasonable prospect of economic extraction are well established.
	☐	All Mineral Resources are reported insitu – exclusive of any modifying factors such as dilution.
	☐	As part of Ore Reserves a 97% mining recovery factor has been applied to surface deposits without large pre-existing depletion voids. This mining recovery factor has been increased in local areas surrounding the historical depletion voids for Gorumbwa and KCD OC. All Mineral Resources are reported insitu.
	☐	KCD underground Mineral Resources for 2017 are insitu Mineral Resources that meet a cut-off of 1.6 g /t Au within a minimum mineable stope shape, reported at and a gold price of $1,500/oz.
	☐	All Mineral Resource tabulations are reported inclusive of that material which is then modified to form Ore Reserves.
Metallurgical Factors or Assumptions	☐	Metallurgical testwork takes into account all of the deposits that are scheduled to be treated in the Kibali Gold Mining Processing Plant (indicated and measured Mineral Resources only).
	☐	The oxide and transition ore flow sheet comprises basic size reduction crushing, milling, gravity and flash flotation followed by carbon in leach process.
	☐	In the case of fresh ore, both open pit and underground, the flow sheet has an additional bulk or rougher flotation stage. Flotation concentrate goes through ultrafine grinding and intensive cyanide leach. This process utilises industry standard technology that has been well tested.
	☐	Metallurgical recovery is mineralisation dependant. KCD oxide, transition, underground fresh, and open pit fresh ores have been declared as 89.9%, 86.1%, 90.1%, and 85.8% respectively. No bulk sample or pilot testwork has been completed; however, the plant has achieved these recoveries on dedicated ore types.
	☐	Each Mineral Resource undergoes independent metallurgical testwork as part of the prerequisites for conversion to reserve as outlined below.
	☐	Mineralogical (AMTEL) examination and BRT testwork on Gorumbwa has identified pockets of samples that gave direct cyanide leachable gold below 80%, where the recoveries can be as low as 54%. However, the bulk of the samples tested gave recoveries in excess of 90% (BRT) and above 80% as predicted by mineralogical examination.

18^th September 2018 Page 360

Kibali Gold Mine NI 43-101 Technical Report

Criteria	Commentary
	☐	Samples were submitted classified into high and low arsenic domain samples. Gold was predominantly found in the coarser size fractions of 75 – 106 µm, invariably associated with the sulphide minerals of arsenopyrite and pyrite in the high arsenic samples while in a finer fraction of 45 -75 µm fraction in the lower arsenic samples. This could be beneficially in flotation recoveries, where a coarser grind leading to better throughputs at mills will favour the running of mills and bulk flotation performance. The understanding of gold occurrence in the sulphides and non-sulphide minerals is critical as this defines the terminal residue in the current Kibali processing flowsheet.
	☐	High Arsenic and low recovery was identified in metallurgical testwork at Sessenge based on an approximate arsenic cut-off of 2,000 ppm. Subsequent BRT samples and arsenic analysis are being completed and applied into the model.
Environmental Factors or Assumptions	☐	All environmental permits are in place for the Kibali processing plant and the KCD underground mine. An environment and social impact assessment was completed as part of the optimised feasibility study in 2012; various other ESIAs were completed for new project elements and these studies were consolidated in 2016 to identify combined impacts.
	☐	Kibali operations have an environmental management program in place, are ISO 14001:2015 compliant and are independently audited to continuously improve environmental management. Audits are also carried out to assess compliance with the International Cyanide Management Code.
	☐	Waste dumps and tailings storage facilities are designed in accordance with good international industry practice. The lined CTSF will need to be enlarged to accommodate increased mine life; the ESIA and associated permitting for this will take place in 2018. No delays are foreseen.
	☐	There were no major or category “A” environmental incidents during the year.
Bulk Density	☐	Specific Gravity values are measured from diamond drill cores by applying the Archimedean principles (density = weight (in air) ÷ (weight (in air) – weight (in water)).
	☐	Density samples are collected on a routine basis from both mineralised zones and ‘waste’ rock. The resultant density distributions are analysed and outliers were removed where appropriate,
	☐	Density for all Kibali Mineral Resources was assigned into the block model based upon weathering, rock unit and mineralisation zones.
	☐	Some zones had insufficient data to ascertain a mean density and thus have an assigned mean density from similar units based upon weathering, rock unit and mineralisation zones.
Classification	☐	Resource Classification was based on geological continuity and data density as well as estimation quality in form of slope of regression (SR) and kriging efficiency (KE). In terms of drill spacing, the following parameters have been applied:
		Deposit	Measured	Indicated	Inferred
	Minimum Samples		8	6	4
	Minimum Consecutive Sections		4	Good Geological Continuity	-

18^th September 2018 Page 361

Kibali Gold Mine NI 43-101 Technical Report

Criteria	Commentary
	Maximum Drilling Density	KCD Open Pit	10 m by 5 m or 20 m by 5 m	40 m by 30 m	80 m by 80 m
		KCD Underground	20 m by 10 m	40 m by 40 m	80 m by 80 m
		Pakaka	20 m by 10 m or 20 m by 5 m	40 m by 40 m	80 m by 60 m
		Mengu Hill	10 m by 5 m	30 m by 20 m	80 m by 80 m
		Rhino	10 m by 5 m	30 m by 20 m	80 m by 80 m
		Gorumbwa		30 m by 30 m	80 m by 80 m
		Kombokolo	10 m by 5 m or 10 by 10 m	30 m by 30 m	80 m by 80 m
		Sessenge	10 m by 10 m	40 m by 40m	80m by 80m
		Pamoa	-	40m by 40m	80m by 80m
		Megi	-	30m by 30m	80m by 80m
		Marakeke	-		80m by 80m
		Mengu village	-	40m by 50m	80m by 80m or 120m by 100m
Audits or Reviews	☐	An independent audit was undertaken in 2012 on the Mineral Resource estimate by Quantitative Group (QG). This focussed primarily on KCD due to its dominant size. The results of which set out some minor recommendations to improve QA/QC compliance, sampling procedures and modelling methodologies which have been acted upon.
	☐	However, the audit failed to identify the bias in drill direction of the sub vertical underground lodes and as such the risk to the resource model was not fully appreciated until 2016 when first advanced GC results drilled from underground development were significantly different to that which had been modelled in the block model. Subsequently, an additional full resource audit Optiro was completed in 2017, after the majority of the underground lodes were covered with an initial pass of resource definition drilling.
	☐	The Mineral Resource estimation processes used by Kibali Goldmines are considered by Optiro to be at a level commensurate with industry best practice based on the process review completed in August 2017 (Optiro, 2017)
	☐	No essential issues were found, but series of recommended issues were raised and to be implemented before the next audit.
	☐	Increase the use of implicit modelling software for lithology and structural modelling.
	☐	Generation of more formal guidelines for the relative interaction of top cutting and high yield restriction
	☐	Value added issues – all of which are planned to be implemented during 2018.
	☐	Inclusion of bias method test were applicable between RC and DD (QQ Plot).
	☐	Trial of Vulcan dynamic anisotropy to deal with more tightly folded portions of Mineral Resource models.
	☐	Use of global change of support validation for the model tonnage-grade curves
Discussion of Relative	☐	The application of optimised mineable resource shapes, applies reasonable mineability constraints including a minimum mining width, a reasonable distance from current or planned development, and a measure of assumed profitability at the related resource cut-off grade. This change in reporting method has removed isolated areas of mineralisation and lowering the grade of the reported underground

18^th September 2018 Page 362

Kibali Gold Mine NI 43-101 Technical Report

Criteria	Commentary
Accuracy/Confidence		resource by reporting all material, geologically classified as ore within each mineable shape, whilst ensuring the overall shape meets the resource cut-off grade. Thereby ensuring that the Mineral Resources are reported in line with industry best practise with specific regard to underground Mineral Resources only being reported if there is an intention to mine the material.
	☐	2017, Optiro completed a resource audit at Kibali which included KCD underground model in line with comments and recommendations in the 2016 resource report. This is due to the substantial advanced grade control drilling completed to understand the bias in drill direction of the sub vertical underground lodes.
	☐	Optiro acknowledges that the estimation of resources at Kibali is complex, with a number of very large models. Kibali Goldmines has tackled the estimation in a systematic way, with largely common approaches to compositing, top cuts, declustering, KNA, estimation parameters, classification and validation. The documentation of the estimation and validation is generally very comprehensive. The processes follow good to best industry practice.

18^th September 2018 Page 363

Kibali Gold Mine NI 43-101 Technical Report

Section 4. Estimation and Reporting of Ore Reserves

As required by Section 4 of the 2012 JORC Code for reporting of Ore Reserves, a table 1 is provided that contains a summary of the estimation and reporting criteria relevant to the Ore Reserve declared.

Criteria	Commentary
	☐	The Kibali KCD underground 31st December 2017 Mineral Resources are the basis of the Ore Reserves estimate. The Ore Reserves are based on geological block models produced in September 2017 (kcd_res_2017_09_25_rev5.bmf).
Mineral Resource Estimate for Conversion to Ore Reserves	☐	In 2017 infill grade control drilling has led to reinterpretations of mineralised zones. The changes included in the September geological block model have been addressed in this Ore Reserve estimate. A later geological block model (November 2017 - KCD_GC_2017_11_25.bmf) included a reinterpretation of the lower 5101 zone. This change has been partially addressed by excluding northern stopes on 240 level in 5101, removing 120 koz Au from the Ore Reserve.
	☐	Mineral Resources are reported inclusive of Ore Reserves.
	☐	The Ore Reserve statement is based upon the Mineral Resource declared as of 31st December 2017 by Randgold Resources Ltd. Mineral Resources are reported inclusive of Ore Reserves.
Site Visits	☐	A site visit was undertaken by the Competent Person - Andrew Law in July 2017, and December 2017. These site visits included a review, coaching and support of mine planning.
Study Status	☐	The study to convert Mineral Resources to Ore Reserves is an operational life of mine plan update. An Optimised Feasibility Study (OFS) was undertaken in 2011-2102. Since that study technical reports and modelling updates have been undertaken. These updates are addressed in stand-alone reports and have not been complied into a study document. The Competent Person has reviewed studies and operational history that support all material Modifying Factors and considers it is at least equivalent to Pre-Feasibility Study level.
	☐	The Ore Reserve is predicated upon a Feasibility Study commissioned by Kibali Goldmines and completed in December 2012. The finding of the Feasibility Study was an economically viable mining operation. The operation poured first gold at the end of 2013 and has been in continuous operation since then. The operation has continued to be optimised since the feasibility. Operational readiness and preparedness are organised for all the pits before they are brought into production to close any gaps from previous learnings on any of the pits.
Cut-Off Parameters	☐	The cut-off grade values are calculated on the basis of mined gold grade, including dilution, in grams per tonne. No by-product credits or metal equivalents are used. Revenue calculation is based on a gold price in $ and takes into account processing losses. Costs are based on 2017 actual operational costs in $.
	☐	For the majority of the Kibali underground Ore Reserve a cut-off grade of 2.5 g/t Au has been used. In the 9105 lode, the cut-off grade has been increased to 2.8 g/t Au. These cut-off grades were applied to stope panels after dilution and ore loss had been accounted for in the panel.
	☐	The cut-off grades have been estimated for each material type for all six reserve pits included in the 2017 Ore Reserve estimate. These are based on a gold price of $1,000/oz and $1,100/oz for the KCD PB3 pit and include dilution, royalties, processing cost and recoveries,

18^th September 2018 Page 364

Kibali Gold Mine NI 43-101 Technical Report

Criteria	Commentary
		general and administration cost, and ore mining costs.
	☐	Cut-off grade analysis for the various deposits and material types are updated each year as per the budgeting of the operation and economic costing.
	☐	The Ore Reserves have been estimated using three dimensional mine designs. These have been created using three dimensional geological block models and wireframes; and take into account the geotechnical environment.
	☐	Open pit mining takes place in a number of satellite pits over approximately 14 km. Some of the pits are relatively shallow and have a short mine life of two years or less such as Pamao and Sessenge, whilst others are deeper and have a longer life of more than two years, such as Pakaka and Gorumbwa. There are six main open pit deposits, KCD, Pakaka, Pamao, Kombokolo, Sessenge, and Mengu Hill, located within an approximately 7 km radius.
	☐	Open pit mining is conducted by contractor Kibali Mining Services (KMS), a local subsidiary of DTP Terrassement, using either free-dig or conventional drill, blast, load, and haul methods. The mining equipment is jointly owned by a subsidiary of Randgold and the contractor’s parent, which also operates at Randgold’s Loulo-Gounkoto Mine in Mali and Tongon Mine in Côte d’Ivoire.
	☐	The upper levels of the open pits are usually in weathered material, which typically is free digging material. Once fresh (unweathered) rock is encountered, drilling and blasting is required. Emulsion explosives are supplied as a down-the-hole service by Orica.
	☐	Free digging in the upper levels uses 5 m high benches, with 10 m benches used for drilling and blasting operations. The 10 m benches containing ore are excavated in three flitches of equal height.
Mining Factors or Assumptions	☐	The Dilution factor for open pit Ore Reserves has been applied based on the nature of the deposit and mining equipment selected. A 10% dilution factor is applied to most open pit Ore Reserves except the areas immediately around known voids in Gorumbwa and KCD. Dilution, when mining around voids has been included at 13%, while dilution at 9.6% has been included when mining the larger lodes. The average dilution can be expected to decrease when the void areas are mined out and mining takes place in normal mining conditions without voids. A 10% dilution allowance has therefore been adopted for the entire KCD deposit.
	☐	Ore loss has been estimated at 92% for the larger void areas in the KCD pit; the ore is expected to be able to be retrieved from lower parts of the voids and so can be considered a delayed recovery rather than total ore loss. A global 97% mining recovery factor has been applied to all mining recoveries in the estimation of open pit Ore Reserves. This has proven to be accurate on a global basis while mining the various pits Locally there are areas that return higher and lower than 10%.
	☐	The minimum mining width is 40 m and the minimum pushback width is 60 m.
	☐	The Ore Reserves have been estimated using three dimensional mine designs. These have been created using three dimensional geological block models and wireframes; and take into account the geotechnical environment.
	☐	The mining method used is long hole open stoping. Predominantly primary / secondary multi lift stoping (~68% of ounces). In the 9101 zone transverse advancing face multi-level stoping is used (~23% of ounces). Longitudinal hole bench stoping is used in narrower stope areas (~9% of ounces). Level intervals vary from 25 m to 35 m. Stopes lengths along strike vary from 20 m (primary), 25 m (advancing face), 30 m (secondary) and up to 50 m for bench stopes. Stope widths for transverse stopes vary up to 40 m, longitudinal stope widths range from 5

18^th September 2018 Page 365

Kibali Gold Mine NI 43-101 Technical Report

Criteria	Commentary
		m to 20 m. Stopes spans are designed to be stable based on a mining rock mass model and stability graph analysis.
	☐	Long hole stoping methods are currently used at KCD UG. Mining factors are supported by reconciliation of previously mined areas, however the previous Ore Reserve ore loss and dilution targets are not being achieved. Dilution external to stope shapes (unplanned dilution) has been increased from 1.8% to 4.8% overall, which is consistent with 2017 performance. External and paste dilution is applied at a grade of 0 g/t. Internal (planned) dilution is at geological block model grade.
	☐	Ore loss has been kept at 3% for most stopes as the stoping to date in the 51015 and upper 5101 zones has been predominantly single lift and is not representative of stoping in the larger zones, 5102 and lower 5101.
	☐	Ore loss in some secondary stopes in 5102 zone has been increased to 10% due to changes in the mining sequence. The mining sequence in 5102 and lower 5101 has been modified to bring forward the 5102 stoping. Some 5102 secondary stopes are expected to have higher ore losses due to higher mining stresses in these stopes and deterioration of the adjacent fill masses. Mine wide three dimensional non-linear modelling is being undertaken to assess the impact of this change. The proposed mining methods are variants of long hole open stoping with cemented paste:
	☐	No significant failures of the openings in the underground workings have occurred. The rock assessed for the rock mass model is ranked as good to very good.
		The underground mining operations are currently operated by contractors (Byrnecut) and the contractor operation will continue until approximately mid to late 2018, when the changeover to The Kibali KCD underground mine has a scheduled production rate of 3.6 Mtpa for 10 years.
	☐	The Kibali gold processing plant comprises two largely independent processing circuits, the first one designed for oxide and transition ores and the second for sulphide refractory ore. However, both circuits are designed to process sulphide ore when the oxide ore and transition ore is no longer available. The flow sheet, comprises crushing, ball milling, classification, gravity recovery, a conventional CIL circuit, flash flotation, also conventional flotation, together producing a concentrate which goes to ultra-fine-grinding and a dedicated intensive cyanide leach. This process consists of well tested technology in the gold industry and is appropriate for Kibali’s style of mineralisation.
Metallurgical Factors or Assumptions	☐	The extensive metallurgical testwork campaigns conducted for Kibali demonstrate two distinct behavioural patterns where some ore types exhibit free-milling characteristics suitable for gold extraction by a conventional carbon in leach (CIL) metallurgical process. Other ore types exhibit a degree of refractoriness, where straight cyanidation returns gold dissolutions considered to be too low for optimal plant operation. The reason for this refractoriness is due to the presence of occluded gold particles within sulphide minerals. It has been demonstrated that a finer grind will expose a portion of this additional gold for leaching so that the recovery is enhanced to economically acceptable levels.
	☐	Most of the ore bodies contain some extent of free native gold, which means it is large enough to recover via a density separating step which is performed with Knelson gravity concentrators during the milling cycle.
	☐	The processing plant rated throughput is 3.6 Mtpa of soft oxide rock ore through the oxide circuit and 3.6 Mtpa of primary sulphide rock ore through a parallel sulphide circuit. Once the plant is sulphide only, the capacity is 7.2 Mtpa of sulphide ore. Kibali’s operational performance has demonstrated that the process plant is fully capable of exceeding its design capacity.

18^th September 2018 Page 366

Kibali Gold Mine NI 43-101 Technical Report

Criteria	Commentary
	☐	The underground ore feed is blended with ore from open pits at Kibali. The process plant has been treating Kibali KCD underground ore since 2015 and has demonstrated consistent recovery performance. Overall, the actual process plant gold recovery in 2017 varied monthly from 80.2% to 85.6%. The average gold recovery in 2017 was 83.4%. Recovery for 2018 is expected to be 84%, increasing to 87% in 2019, and 89% in subsequent years based primarily on a shift away from a blended ore feed to one that will be dominated by the KCD and better recovery ores namely Gorumbwa deposit.
	☐	Metallurgical testwork program undertaken across the deposits at feasibility study stage is representative of the current Ore Reserve. Ongoing testwork is undertaken regularly on samples from production areas.
	☐	Fresh ore samples, comprising both underground and open pit, from KCD lodes have been selected to ensure representivity of the fresh ore of the KCD deposit. Metallurgical domains of the fresh underground ore were defined between the 3 lodes tested namely the 3000, 5000 and 9000 lodes
	☐	Additional testwork has been conducted by Kibali Mines on all subsequent satellite pits that have been brought into production. The different deposits respond differently and a range of recoveries from 75% to 90% are achieved from the different mineralisation.
	☐	Metallurgical recovery of oxide, transition, KCD open pit fresh and KCD underground ores have been declared as 89.9%, 86.1%, 85.8% and 89.9% respectively. Gold recovery averaged 80% on the plant in 2016, through a mixture of KCD open pit and underground, Pakaka, Mengu Hill and Kombokolo Metallurgical recoveries are expected to improve as the component of satellite ore reduces
	☐	The main deleterious element in the Kibali ore sources is considered to be arsenic. Certain isolated ore types exhibit higher levels of arsenic (in Pakaka and Sessenge) which can result in dissolution during the recovery process. The impact of arsenic was in the leach of flotation concentrate in the intensive oxygenation/cyanidation circuit. Arsenic content in excess of 2,000 ppm has a negative effect on gold dissolution where dissolution values as low as 70% are attained when arsenic content increases to values as high as 9,000 ppm.
	☐	Detailed geometallurgical analysis have been completed on Pakaka and Sessenge where the arsenic content has been modelled as part of the Mineral resource block model. This has enabled the application of a blending strategy where high arsenic content ores are intentionally mined and stockpiled separately such that they can be blended with low arsenic ore sources, thereby restricting the arsenic solution tenors within the circuit.
Environmental	☐	All environmental permits are in place for the Kibali processing plant, KCD pit and underground mine and the Satellite pits. Permits for the hydropower plants have also been received. An environment and social impact assessment was completed as part of the optimised feasibility study in 2010/11; various other ESIAs were completed for new project elements and these studies were consolidated in 2014 and 2105 to identify combined impacts.
	☐	Kibali operations have an environmental management program in place, are ISO 14001:2015 compliant and are independently audited to continuously improve environmental management. Audits are also carried out to assess compliance with the International Cyanide Management Code.
	☐	Waste dumps and tailings storage facilities are designed in accordance with good international industry practice with adequate design

18^th September 2018 Page 367

Kibali Gold Mine NI 43-101 Technical Report

Criteria	Commentary
		capacity for the life of mine. The lined CTSF1 and CTSF2 be incorporated to accommodate increased mine life. There were no major or category “A” environmental incidents during the year.
	☐	The majority of the infrastructure required for mining and processing is already in place. The Life of Mine Plan includes provision of additional infrastructure for mine ventilation and materials handling.
Infrastructure	☐	Surface infrastructure associated with the overall Kibali operation includes a processing plant, tailings storage facility, camp, power stations, airstrip, workshops, offices and all other associated infrastructure required. All of these items have been fully designed, costed and accounted for in the economic assessment of the Project.
	☐	The new cyanide tailings storage facility (CTSF2) was commissioned in January 2016 and will be providing process capacity until 2018. Future LoM CTSF capacity is planned through two facility raises in 2019 and 2023.
	☐	The Ambarau hydropower generation plant was commissioned in 2017 and power generation underway.
	☐	Construction of the Azambi power plant is underway and expected to be commissioned by June 2018.
	☐	Operating costs (mining, processing, site general and administration) are derived from 2017 actual costs at Kibali. Life of mine costs have been adjusted to allow for operational changes including changing from contract underground mining to owner operator mining (planned for 2018).
	☐	Development capital costs are derived from mine designs and are costed by allocation of a portion of the total development cost during each period. The allocation cost includes a share of mining overhead costs. Plant and equipment capital costs have been estimated by year for the life of mine.
	☐	Nil allowances were made for deleterious element as there are no significant levels of deleterious elements are present in the gold bullion.
Costs	☐	Transportation of gold bullion from site and refining charges are included in operating costs for the processing plant and are derived from existing agreements.
	☐	A royalty and an export tax which combined is 3.5% of gold sales is payable to the DRC government.
	☐	Costs used in for the pit optimisations were derived from the Mining Contractor’s pricing of the open pit life of Mine schedule. Owners cost were also added.
	☐	Labour costs for national employees were based actual costs. Local labour laws regarding hours of work etc. were also considered and overtime costs included.
	☐	Customs duties, taxes, charges and logistically costs are included in all relevant areas of the Project.
	☐	During 2017 costs for processing and G&A were updated based on actuals adjusted for latest forward estimates, production profiles and manning levels.
Revenue Factors	☐	The assumptions made for commodity prices are: gold $1,000 per troy ounce.
	☐	This value is based on a conservative view of gold price following a period of relatively higher gold price volatility.

18^th September 2018 Page 368

Kibali Gold Mine NI 43-101 Technical Report

Criteria	Commentary
	☐	A gold price of $1,000/ oz was employed for the pits for the Ore Reserve process except for the KCD Pushback 3 where $1,100 /troy oz was used. The KCD open it is however limited at the 5865L open pit – underground interface.
Market Assessment	☐	The gold market is highly liquid and benefits from terminal markets (London, New York, Tokyo, and Hong Kong) on almost a continuous basis. Gold prices were in general on a downward trend from 1980 to 2000 where it traded down to approximately $250/oz. Between 2000 and 2011 the market was on a general upward trend that moved spot prices to a peak of $1,900/oz. 2013 saw a sharp correction in the upward trend, with the spot price dropping to $1250/oz. Since 2014 the gold price, have traded in range of $1,050 to $1,400 / troy. Oz.
	☐	Gold produced at the mine site is shipped from site, under secured conditions, to a refining company. Under pre-established contractual conditions, the refiner purchases the gold from the mine with the proceeds automatically credits the mines’ bank account. The operation is unhedged.
	☐	The economic analysis of the life of mine uses a range of discount rates from 2.5 to 10%. No inflation of costs or revenues has been included in economic analysis. Sensitivity analysis to gold price, mined grade and operating costs have been undertaken. The economic outcome for the KCD underground is fairly insensitive to likely variation in economic variables. Only 10% of the Ore Reserve is between 2.50 g/t Au and 3.5 g/t Au. The cut-off grade is based on a gold price of $1,000 / per ounce, current spot gold prices are ~20% above this price.
	☐	Gold price: $1,000 flat
	☐	Royalty: 3.5% - 2.5% royalty and 1% selling cost
Economic	☐	Income tax: 30%, no tax holiday, but accelerated depreciation incorporated
	☐	Met Recovery: 89%
	☐	Open pit mining cost: $3.27/t mined
	☐	Underground mining cost: $34.46/t ore mined
	☐	Processing: $17.20/t processed
	☐	G&A: $7.78/t processed
	☐	LOM Capital: $371 M including construction, ongoing, UG capital development and drilling, pre-production, exploration, rehab and closure
Social	☐	The mine has not faced any material social issues. However, there is a growing need to reinforce communication and sensitisation in the community regarding mining and its benefits. The mine will continue its engagement to manage this potential risk.
	☐	The various ESIAs carried out for Kibali and associated infrastructure have included assessments of impacts and benefits to local communities.
	☐	Stakeholder engagement and dialogue is ongoing and the importance of this to the company is demonstrated by the fact that senior managers have KPIs related to their involvement in dialogue with communities, which generally requires them to attend quarterly

18^th September 2018 Page 369

Kibali Gold Mine NI 43-101 Technical Report

Criteria	Commentary
		stakeholder meetings. The mine also has a separate Social and Community Department to focus on this stakeholder engagement.
	☐	A grievance mechanism is in place.
	☐	The mine is a significant employer to members of the local communities. The underground mining operations contribute to extended life-of- mine, employment of local Congolese and the growth of the DRC economy. Kibali Goldmines policy is to promote nationals to manage the project. Where locally qualified and experienced staff is not available, recruitment from elsewhere is undertaken, with the clear understanding that local personnel are given the training and experience required to allow them to replace the expatriates as soon as possible.
	☐	Kibali Goldmines policy of promoting local employment also extends to its contractors. 2017 figures show that the workforce was made up of 83% contractors, of whom 92% were nationals, and 17% employees, of whom 88% were nationals. Overall local employees number 4,917 (92%), out of a total of 5,377 employer and contractor posts.
	☐	Local procurement is also promoted and is a requirement of contractors as well as Kibali Goldmines. Where possible, goods and services are procured locally. This includes produce from the various agribusinesses (eggs, pork, maize) which is purchased for use in the mine canteens.
	☐	Two significant resettlement campaigns have taken place, one in 2012/2013 and one in 2016/2017. Ongoing monitoring of affected households to ensure that their livelihoods, often previously based on artisanal mining, are not adversely affected by the resettlement, will be ongoing. Economic displacement has also been significant across the area.
	☐	Stakeholder engagement activities, community development projects and local economic development initiatives contribute to the maintenance and strengthening of Kibali’s Social License to Operate (SLTO). This includes regional as well as local engagement using a variety of means include radio broadcasts. The collaboration between Kibali Goldmines and the community keeps improving as discussions are more constructive and participative.
	☐	Kibali Goldmines continues to invest in community development initiatives, focussing on potable water supplies, primary school education, health care education, investment in medical clinics and local economic development projects; these, and livelihood projects, such as the program to improve the agricultural yield of the area should be continued.
	☐	The ongoing presence of artisanal miners operating in and around the Permit area has the potential to cause unrest. There are plans in place to develop a cessation strategy and engage with the Haut Uélé governor for his full involvement in the cessation process, as well as sensitising the local community and local chiefs to continue ASM activities in the ‘corridors’ identified for ASM by the government.
	☐	The DRC Mining Code (2002) and Regulations have been amended with an updated Mining Code which came into force on 9th March 2018 (DRC Mining Code(2018)) and the related amended Mining Regulations which came into force on 8th June 2018.
Other	☐	The following changes have been made to the DRC Mining Code (2002) that could have an impact on Kibali:
		o   Royalty charges are to be increased from 2.5% to 3.5%. This increases royalty charges over the LOM by an estimated $94.5 M, which would not materially impact the LOM profitability.
		o   Various increases in import and other duties from 4% to 7% depending on consumable type, which would not materially impact the

18^th September 2018 Page 370

Kibali Gold Mine NI 43-101 Technical Report

Criteria	Commentary
		LOM profitability.
		o   A super-tax profit has been promulgated based on the feasibility study prepared at the time the approval was given for the building of the Project. Accordingly, such a tax would only apply if the average annual gold price was in excess of $2,000/oz.
	☐	The exact impact, if any, of the changes will only be fully known once the 2018 Mining Code and related regulations are clarified and implemented in full.
	☐	Going forward the DRC Mining Code (2018) envisages a stability period for the tax, customs and exchange control regime of five years from the date on which the DRC Mining Code (2018) came into force and further provides that a number of the taxes shall be applied in accordance with the applicable substantive law.
	☐	Kibali Jersey Limited, the holding company of Kibali, the shareholders of Kibali Jersey Limited and Kibali Gold Mines SA, are considering all options to protect their vested rights under the DRC Mining Code and to enforce the additional state guarantees previously received, including preparations for international arbitration. In addition, engagement with the DRC government is ongoing, with the aim of exploring alternative solutions, which could be mutually acceptable to both parties. This includes the application of Article 220 of the DRC Mining Code (2018), which affords benefits to mining companies such as Kibali, operating in landlocked infrastructurally challenged provinces. If Article 220 were applied to Kibali, any advantages granted would mitigate any impact of the implementation of the DRC Mining Code (2018).
	☐	The mining code is in the process of transition and the current proposed changes do not have an impact on the stated Mineral Resource or Ore Reserves at the gold prices used for Ore Reserve ($1,000/oz), Mineral Resource ($1,500/oz), or the current gold prices of $1,200/oz (August 2018).
Classification	☐	Kibali Ore Reserves are classified as Proved and Probable Ore Reserves and are based on confidence levels determined in the Mineral Resource. Measured Mineral Resource has been reported as Proved Ore Reserve.
	☐	The Kibali KCD UG Ore Reserve is classified as 28% Proved and 72% Probable Ore Reserve. The Ore Reserve classification result of KCD UG Proved Ore Reserves being declared for the first time in 2017 reflects The change is due to substantial infill grade control drilling and ore development providing exposures for mapping in the areas of Measured Mineral Resource; and processing of 4.3 Mt of underground ore to date providing higher confidence in mining and processing modifying factors.
	☐	Stockpiles on the surface are classified as Measured Mineral Resource and Proved Ore Reserve. These stockpiles are reported as part of the Kibali Open Pit Ore Reserves estimate, not this estimate.
Audits or Reviews	☐	In 2013 the Kibali Mineral Resources were reviewed by QG Consulting (QG). QG concluded that Mineral Resources are likely to be free of material error. Recommendations were made on some points as opportunities improvement in processes and understanding. These recommendations have been addressed.
	☐	In 2014 Snowden Mining Industry Consultants (Snowden) audited the 2013 Kibali Ore Reserve estimates. Snowden concluded that the Ore Reserves are in accordance with JORC Code (2102) and other relevant international reporting codes. Recommendation were made on some points for improvement related to the KCD underground Ore Reserves. These recommendations have been addressed.
	☐	In 2017 Optiro Pty Ltd (Optiro) audited the 2017 Kibali Ore Reserve estimate (including this estimate). The results of the audit show that

18^th September 2018 Page 371

Kibali Gold Mine NI 43-101 Technical Report

Criteria	Commentary
		Ore Reserve process areas reviewed correspond to good practice or best practice. Final audit recommendations are being addressed by Kibali.
	☐	As part of the Feasibility Process – regular formal reviews were held with the Kibali JV technical representatives.
Discussion of Relative Accuracy/ Confidence	☐	Accuracy and confidence level in the Ore Reserve estimate has been assessed qualitatively.
	☐	On a global scale in 2017 additional infill grade control drilling and reinterpretation of geological setting has led to a slight reduction in Ore Reserves in existing areas (excluding new areas converted from Inferred Mineral Resources).
	☐	On a local scale there have been instances of reduction or gains in parts of the orebodies.
	☐	Opportunities exist with the Inferred Mineral Resource within the current pits that can be upgraded and converted to Ore Reserve with drilling.
	☐	The down plunge continuity of the Ore Reserves across both open pit and UG continues to supply additional opportunity for extensions of resources and reserves while the short range variation across strike is often the reason for local changes to interpretation. Continued infill drilling and exploration drill programs are required to continually identify and measure the risks and opportunities. Since 2016 significant drill programs have been put in place.
	☐	The mining sequence in the 5102 and lower 5101 zones has been modified to bring forward the 5102 stoping. The impact of this change is uncertain and modelling is underway to assess this impact. An increased ore loss (10%) has been applied to some 5102 secondary stopes due to likely higher mining stresses in these stopes and deterioration of the adjacent fill masses.
	☐	The modifying factors used are considered realistic to conservative for this project based on the geotechnical environment and mining methods selected. The values applied are in-line with published data from similar scale operations.

18^th September 2018 Page 372