The output of agricultural nutritive substance is the one of main cause for water environment pollution. Taking Miyun reservoir in Beijing as study area, combined with the land use, soil and precipitation data, under the support of the GIS and RS technology, the long-term average pollution loads in study area were estimated using the Long-Term Hydrologic Impact Assessment(L-THIA) model. The spatial distribution of the pollution loads was analyzed. Results indicate that the pollution loads are different under different land use types, and the agricultural land has the most important impact on non-point pollution in the study area. The result of this research shows that L-THIA model is available for the spatial distribution of non-point pollution and it can provide the service for planning and management of the watershed.