Abstract: In order to find the best interpolation method of SOM distribution in the Mollisol watershed, 3S combined with geostatistics was used to extract topographic factors and analyze SOM distribution at the soil depth of 0-20 cm in the Haigouhe watershed of Heilongjiang province. CK (Co-Kriging), RK (Regression Kriging) and GWRK (Geographically Weighted Regression Kriging) were adopted to predict SOM distribution, and the accuracy of the interpolation was also investigated in the present study. The main results indicated that the mean SOM was 12.96 g/kg at 0-20 cm soil depth for the whole watershed, with a degree of variation at a moderate level. Altitude and distance from the river system were significantly correlated to SOM, and could be used as cofactors for SOM interpolation. Compared to CK, the accuracy of interpolation increased to 56.09% and 90.87% for RK and GWRK, respectively. The residual from GWR was calculated and interpolated, and also improved the accuracy of prediction. Furthermore, the accuracy of the prediction could also be improved effectively by including anthropomorphic factors into the interpolation; the method should be adopted in future research work. Therefore, GWR could be recognized as the best interpolation method, but could be improved in the future.