Net primary productivity (NPP) is a critical indicator of carbon stocks. Understanding its geospatial drivers is essential for regional carbon cycle research. This study analyzed the spatial and temporal evolution characteristics of vegetation NPP in Heilongjiang Province from 2000 to 2020 using trend analysis based on MOD17A3 data. Subsequently, the spatial differentiation of the factors driving changes in NPP was explored using the multi-scale geographically weighted regression (MGWR) model, a novel spatial statistical analysis method. The results indicated that NPP showed an increasing trend in most regions of Heilongjiang Province over the past two decades. The average value of NPP increased from 348.90 g C m^-2 a^-1 to 454.00 g C m^-2 a^-1, reflecting a growth rate of 29.95%. Significant increases and decreases in NPP primarily occurred in regions with noticeable improvement in vegetation cover, as well as intensive expansion of arable land and urbanization. The MGWR model performed well overall, with an adjusted R² value of 0.875. The model bandwidth revealed that precipitation, temperature, population density, and land use change contributed to NPP at the county scale, while road density contributed to NPP at the city scale. The response of vegetation NPP to these driving factors varied notably across regions. In the Xinganling Mountains, NPP was influenced by land use change and climate factors, whereas in the Sanjiang Plain and Songnen Plain, land use change was the dominant factor affecting NPP. Land use changes driven by ecological protection and restoration initiatives, as well as the expansion of farmland and urban areas, were important contributors to the changes in vegetation NPP in Heilongjiang Province. This study improved the understanding of vegetation dynamics and its driving mechanism in Northeast China, providing a scientific basis for improving the ecosystem carbon sink.