Filed by Automated Filing Services Inc. (604) 609-0244 - NovaGold Resources Inc.

A review was conducted of the 2006 and 2007 Donlin Creek QAQC Program which included blanks, standards and duplicate samples. Data used in this review was provided by Penny Hobby and Rich Harris of Barrick Gold Corp. in January 2007. An insertion rate of one in twenty for blanks, standards and duplicates was quoted in the report “Sampling Method and Approach (after the 43-101 report)”; however, this cannot be confirmed as results for non-control samples were not provided in the data package.

Blank data was provided within the file QAQC_STD_070128_pch.csv. Two types of blank material were used during the program; a product (Blnk) from Anchorage Sand and Gravel for the beginning of the 2006 season, followed by granite chips (Blnk-Gr) purchased from Lowe’s for later 2006 and 2007. Each blank type was analysed for Au by two different assay protocols, Au-AA23 and Au-AA25, and can be divided as follows:

Blnk Type	Total Count
Blnk AA23	36
Blnk AA25	5
Blnk-Gr AA23	1435
Blnk-Gr AA25	2083
Sum	3559

Suite 2300, 200 Granville Street, Vancouver, BC V6C 1S4 Canada
Telephone 604-669-6227 • Facsimile 604-669-6272 • www.novagold.net
Toronto Stock Exchange: NG • American Stock Exchange: NG

According to an excerpt from the report “Sampling Method and Approach (after the 43-101 report)”, acceptable limits for the blank material was typically <0.05ppm Au until late 2006, after which the limit was set at <0.02 ppm Au. For the purpose of this review “Late 2006” was interpreted to mean following a date of September 1, 2006. Blank performance was also reviewed in terms of a more typical upper limit of three times the detection limit in the element of interest.

At an upper limit of three times the detection limit for Au, a total of 246 blanks were found to exceed this, resulting in a failure rate of 6.9% . The detection limits for Au-AA23 and Au-AA25 are 0.005 ppm and 0.01ppm Au, respectively. However, when examining the blanks in terms of the cut-off values outlined by Barrick Gold Corp., the failure rate is more modest. By dividing the samples into two time frames, a total of 5 samples or 0.1% of blanks failed before September 2006 and 163 samples or 4.6% of blanks failed after September 2006. The data that was extracted from the Donlin Creek AcQuire database included comments regarding many of the failed blanks. Of the five failures before September 2006, only one contained a comment. Of the 163 blanks that failed following September 2006, seventy-five had attached comments, leaving eighty-eight without. Examples of such comments are as follows:

‘High Au value for Blank; but within acceptable limits’
‘Failed QAQC for Au by a high margin on original certificate’
‘Blank slightly out of range’
‘Failed QAQC slightly for Blank-Granite, probably due to cross
over contamination from surrounding samples.’

Although these comments do not always explain why the sample was not sent for re-assay, it at least indicates that the failure was noted, and was consciously ignored. A definition of “acceptable limits” as mentioned in the comments above would be helpful in reducing the ambiguity of some of these statements. However, there are still a total of ninety-two blanks (just over 50% of all blank failures), that have no comment and have not been accounted for. The maximum failures were found to be 0.135ppm, 0.438ppm and 0.64ppm for Blnk-AA23, Blnk-Gr AA23 and Blnk-Gr AA25 respectively, and their significance in terms of ore-grade cut-off

should be determined. A number of the blank failures could be a result of smear from proceeding higher-grade samples; however, this could not be investigated as proceeding assay sample data was not provided as part of the data package.

Charts depicting each blank type, further dived by assay protocol and differing detection limits can be seen in Appendix I. Failure lines can be seen at 0.05ppm Au, 0.02ppm Au and three times the detection limit (0.015ppm for Au-AA23 and 0.03ppm for Au-AA25). Please note that these charts would be altered slightly if further separated into the two date ranges of before and after September 1, 2006.

Duplicate data was obtained from the file QAQC_DUP_070128_pch.csv. A total of 3885 duplicate pairs obtained from coarse reject material were taken at Donlin Creek during the 2006 and 2007 programs. Successful or “passing” duplicates were identified by calculating the Absolute Relative Difference (ARD):

For coarse reject duplicates, the following criteria (outlined in ‘Assay Quality Assurance-Quality Control Program for Drilling Projects at the Pre-Feasibility to Feasibility Report Level 5^th Edition’, S. Long, 2007) must be met: ARD must be <0.2 for 90% of the pairs, and if the Pair Mean <15(DET), the Pair Difference must be ≤ 3(DET) to pass. Detection limits are 0.005ppm and 0.01ppm Au for Au-AA23 and Au-AA25 respectively, and results below detection limit were assigned a value equal to half the detection limit, to avoid the ARD equation becoming undefined. After dividing the duplicate pairs by assay method, and applying the ARD equation, the pairs were found to be 93% passing.

Charts were plotted to show the duplicate pairs and their Pass/Fail lines. The max value of each pair was plotted against the min value in order to reduce bias and plot above the y=x line. Ideally, all Donlin Creek duplicates should plot between the y=x line and the 20% Pass/Fail (P/F) line. Pass/Fail lines cross the y-axis at three times the detection limit of the element of interest. Duplicate performance is also represented as an ARD (or AVRD: Absolute Value of the Relative Difference) vs. Cumulative Frequency (in Percentile) plot. Outliers in the duplicate pairs can be visually identified as points that fall to the right of the inflection point, where the slope of the line steepens greatly. For both assay methods, less than 10% of the pairs occur to the right of this point. All plots can be seen in Appendix II.

A total of 3573 results from fifteen different SRMs or standard reference materials were provided in the file QAQC_STD_070128_pch.csv. For each standard, the total number of samples as well as the mean and standard deviation of the returned assay results can be seen in the chart below. The standards were also divided by assay method, and upper and lower performance brackets are provided for both two and three standard deviations from the mean.

Standard	Method	Count	Mean	Std Dev	Mean - 2SD	Mean + 2SD	Mean - 3SD	Mean + 3SD
LS1	Au-AA23	27	0.0049	0.0005	0.0039	0.0059	0.0035	0.0064
Std-G	Au-AA23	686	1.4775	0.0638	1.3500	1.6051	1.2862	1.6688
Std-G	Au-AA25	92	1.4423	0.1427	1.1569	1.7276	1.0142	1.8703
Std-H	Au-AA23	523	2.9535	0.1256	2.7024	3.2046	2.5768	3.3302
Std-H	Au-AA25	138	2.9478	0.1092	2.7293	3.1662	2.6201	3.2754
Std-IC	Au-AA25	121	1.2316	0.0311	1.1693	1.2938	1.1382	1.3249
Std-IH	Au-AA25	44	7.0702	0.2019	6.6665	7.4740	6.4646	7.6759
Std-IL	Au-AA25	132	0.1030	0.0073	0.0884	0.1175	0.0811	0.1248
Std-IM	Au-AA25	146	3.2318	0.6953	1.8413	4.6224	1.1460	5.3177
Std-OX	Au-AA25	28	3.0282	0.0890	2.8502	3.2062	2.7612	3.2952
Std-SC	Au-AA25	119	0.9920	0.2405	0.5110	1.4729	0.2705	1.7134
Std-SH	Au-AA25	29	6.6928	0.1516	6.3895	6.9960	6.2379	7.1476
Std-SL	Au-AA25	716	0.0958	0.0081	0.0795	0.1121	0.0713	0.1202
Std-SM	Au-AA25	72	2.9657	0.4931	1.9795	3.9519	1.4864	4.4450
CDN-GS-3	Au-AA23	251	0.7959	0.0463	0.7032	0.8886	0.6569	0.9349
CDN-GS-3	Au-AA25	441	0.7959	0.0439	0.7081	0.8837	0.6642	0.9276
CDN-GS-3B	Au-AA25	6	3.5183	0.1556	3.2071	3.8296	3.0515	3.9852
CDN-GS-1P5A	Au-AA25	2	1.0650	0.3748	0.3155	1.8145	-0.0593	2.1893

Round-Robin results were provided from Barrick for all standards, excluding CDN-GS-3B, CDN-1P5A and LS1. However, CDN-GS-3B and 1P5A had only six and two samples

respectively. Standard LS1 was used only for limestone drilling which did not support the assays encompassed in the resource estimate. The best value and standard deviation, as well as the number of assay labs participating in the Round-Robin, were used to determine the 95% confidence interval of each standard. Please note that these confidence intervals may vary from those previously provided by Barrick; the confidence intervals for this review were calculated using n=number of labs, not n=number of results, so that the uncertainty in each SRM would reflect the level of disagreement between the labs rather than the differences in analytical precision or how many times the material was assayed. Round-Robin data was also given for both 30 and 50g fire assay methods. Since the assay protocols used for the Donlin Creek data, Au-AA23 and Au-AA25, are both 30g fire-assays, the best values and confidence intervals used for this review are highlighted in yellow. Also note that the best value and standard deviation were not recalculated for the purpose of this study, as individual assay results from the Round-Robin were not provided for the majority of the standards.

Standard	Assay Method	Best Value Au (ppm)	Standard Deviation	Number of Labs	95% C.I.	95% C.I (N=# of labs)
Standard	Assay Method	Provided	Provided	Provided	Provided	Calculated
Std-G	30g FA	1.444	0.093	7	0.18	0.0689
Std-H	30g FA	2.92	0.18	7	0.36	0.1333
Std-IC	30g FA	1.213	0.03	7	NA	0.0222
Std-IH	30g FA	6.929	0.25	7	NA	0.1852
Std-IH	50g FA	6.954	0.29	7	NA	0.2148
Std-IL	30g FA	0.100	0.004	7	NA	0.0030
Std-IM	30g FA	3.124	0.16	7	NA	0.1185
Std-IM	50g FA	3.079	0.12	7	NA	0.0889
Std-OX	30g FA	3.042	0.10	7	NA	0.0741
Std-OX	50g FA	2.943	0.11	7	NA	0.0815
Std-SC	30g FA	0.994	0.04	7	NA	0.0296
Std-SH	30g FA	6.442	0.25	7	NA	0.1852
Std-SH	50g FA	6.496	0.30	7	NA	0.2222
Std-SL	30g FA	0.097	0.01	7	NA	0.0074
Std-SM	30g FA	3.006	0.07	7	NA	0.0519
Std-SM	50g FA	2.981	0.10	7	NA	0.0741
CDN-GS-3	30g FA	0.79	0.037	6	0.07	0.0296
CDN-GS-3B	NA	NA	NA	NA	NA	NA
CDN-GS-1P5A	NA	NA	NA	NA	NA	NA
LS1	NA	NA	NA	NA	NA	NA

A control chart for each standard with sufficient data was created, and can be found in Appendix III. The control charts show lines defined by the best value and the best value ± 5%, which are compared to the lab’s global and moving averages. As per S. Long (Amec Consultant, 2007) the laboratory’s accuracy is considered acceptable if the lab’s moving average stays within ± 5% of the best value. Standard results were also plotted against order of assay to determine any relative bias based on prolonged periods of high or low values. If these periods of high or low values, or “clusters”, are more than 10% above or below the Best Value, the accuracy of the lab during that time frame is off by more than 10%, which is not acceptable for measuring ore grade and these clusters would be flagged with red or blue lines. Results from ore grade samples of the principle commodity should show a low bias; generally speaking, a bias of <5% is widely accepted and a bias of >10% widely rejected.

In 2006, Barrick set acceptable limits for all standards at the mean ± 2 standard deviations. Failed standards were to be re-assayed until the results were acceptable or validated by duplication. The results of the control charts from this review are summarized in the table below:

Standard	Method	Date Range	No. of Outliers	Relative Bias (%)	Time Frames w/ Bias	No. of Samples Exceeding 2SD	No. Exceeding 3SD w/ Comments	No. of Samples Exceeding 3SD	No. Exceeding 3SD w/ Comments	Comments
Std-G	Au-AA23	Jan-Nov 2006	2	1.7	0	14	2	1	0	*Comments required for failed standards.
	Au-AA23	Dec 2006- Feb 2007	3	3.5	0	14	2	1	0	*Comments required for failed standards.
	Au-AA25	NA	2	1.4	0	8	5	4	4	*3 outliers removed, obvious std swap.
Std-H	Au-AA23	2006	3	1.0	0	28	2	5	1	*Comments required for failed standards.
	Au-AA23	2007	0	3.4	0	28	2	5	1	*Comments required for failed standards.
	Au-AA25	NA	2	1.0	0	5	1	1	1	*Comments required for failed standards.
Std-IC	Au-AA25	NA	0	1.5	0	5	5	0	0	*Comments required for failed standards.
Std-IH	Au-AA25	NA	0	2.0	0	3	0	0	0	*Comments required for failed standards.
Std-IL	Au-AA25	NA	1	3.0	1	6	6	0	0	*Comments required for failed standards.
Std-IM	Au-AA25	NA	2	0.4	0	3	3	0	0	*3 outliers removed, obvious std swap.
Std-OX	Au-AA25	NA	0	-0.5	0	1	0	0	0	*Comments required for failed standards.
Std-SC	Au-AA25	NA	0	-0.8	0	3	3	3	3
Std-SH	Au-AA25	NA	0	3.9	0	0	0	0	0
Std-SL	Au-AA25	Jan-Aug 2007	7	-2.2	0	16	10	5	5	*Comments required for failed standards.
Std-SL	Au-AA25	Sept-Dec 2007	3	-0.4	0	16	10	5	5	*Comments required for failed standards.
Std-SM	Au-AA25	NA	0	1.4	0	2	2	2	2	*2 outliers removed, obvious std swap.
CDN-GS-3	Au-AA23	NA	0	0.7	0	11	5	0	0	*Comments required for failed standards.
CDN-GS-3	Au-AA25	NA	0	0.9	0	18	14	0	0	*Comments required for failed standards.
CDN-GS-3B	Au-AA25	NA	NA	NA	NA	NA	NA	NA	NA	*Only 6 samples.
CDN-GS-1P5A	Au-AA25	NA	NA	NA	NA	NA	NA	NA	NA	*Only 2 samples.
LS1	Au-AA23	NA	NA	NA	NA	NA	NA	NA	NA	* Limestone Drilling only, not included in resource model.

Factors that warrant further investigation or are cause for concern are highlighted in pink. All remaining outliers should be examined and determined whether or not they are a logical result of a standard swap, etc. If so, these outliers should be removed from the data set and the control charts should be rerun. A total of 123 standards exceeded the range of ± two standard deviations; fifty-eight (or 47%) were flagged with a comment regarding the failure. These comments commonly referred to a standard swap, or that the result failed by “X %” but was “within acceptable limits”. Again, a definition for “acceptable limits” should be determined to remove ambiguity. The other sixty-five failed standards should be re-examined and flagged with appropriate comments. All failures that cannot be explained logically, are not within “acceptable limits”, or do not fall within material defined as “waste” should be sent for re-assay. If one looks at the performance of the SRMs in terms of three standard deviations, instead of two, only twenty-one samples exceed the tolerated range, sixteen of which have an applicable comment. This leaves only five samples to be re-examined.

To:	Kevin Francis
From:	Melissa Zack
Date:	February 1, 2008
Subject:	Donlin Creek 2006/2007 QAQC

Method	DET	Total Pairs	Pairs Passing	% Passing
Au-AA23	0.005	1474	1371	93.0
Au-AA25	0.01	2411	2249	93.3