Heilongjiang province is the largest forest zone in China and the forest coverage rate is 46%. Forests of Heilongjiang province play an important role in the forest ecosystem of China. In this study we investigated the spatial distribution of forest carbon storage in Heilongjiang province using 3083 plots sampled in 2010. We attempted to fit two global models, ordinary least squares model(OLS), linear mixed model(LMM), and a local model, geographically weighted regression model(GWR), to the relationship between forest carbon content and stand, environment, and climate factors. Five predictors significantly affected forest carbon storage and spatial distribution, viz. average diameter of stand(DBH), number of trees per hectare(TPH), elevation(Elev), slope(Slope) and the product of precipitation and temperature(Rain_Temp). The GWR model outperformed the two global models in both model fitting and prediction because it successfully reduced both spatial autocorrelation and heterogeneity in model residuals. More importantly, the GWR model provided localized model coefficients for each location in the study area, which allowed us to evaluate the influences of local stand conditions and topographic features on tree and stand growth, and forest carbon stock. It also helped us to better understand the impacts of silvicultural and management activities on the amount and changes of forest carbon storage across the province. The detailed information can be readily incorporated with the mapping ability of GIS software to provide excellent tools for assessing the distribution and dynamics of the forest-carbon stock in the next few years. Heilongjiang province is the largest forest zone in China and the forest coverage rate is 46%. Forests of Heilongjiang province play an important role in the forest ecosystem of China. In this study we investigated the spatial distribution of forest carbon storage in Heilongjiang province using 3083 plots sampled in 2010. We attempted to fit two global models, ordinary least squares model (OLS), linear mixed model (LMM), and a local model, geographically weighted regression model (GWR), to the relationship between forest carbon content and stand average diameter of stand (DBH), number of trees per hectare (TPH), elevation (Elev), slope (Slope) and the product of precipitation and temperature (Rain_Temp). The GWR model outperformed the two global models in both model fitting and prediction because it successfully reduced both spatial autocorrelation and heterogeneity in model residuals. More importantly, the GWR model provided localized model coefficients for each location in the study area, which allowed us to evaluate the influences of local stand conditions and topographic features on tree and stand growth, and forest carbon stock. It also helped us to better understand the impacts of silvicultural and management activities on the amount and changes of forest carbon storage across the province. The detailed information can be readily incorporated with the mapping ability of GIS software to provide excellent tools for assessing the distribution and dynamics of the forest. -carbon stock in the next few years.