The Revised Universal Soil Loss Equation(RUSLE) was applied to assess the spatial distribution and dynamic properties of soil loss with geographic information system(GIS) and remote sensing(RS) technologies.To improve the accuracy of soil-erosion estimates,a new C-factor estimation model was developed based on land cover and time series normalized difference vegetation index(NDVI) datasets.The new C-factor was then applied in the RUSLE to integrate rainfall,soil,vegetation,and topography data of different periods,and thus monitor the distribution of soil erosion patterns and their dynamics during a 30-year period of the upstream watershed of Miyun Reservoir(UWMR),China.The results showed that the new C-factor estimation method,which considers land cover status and dynamics,and explicitly incorporates within-land cover variability,was more rational,quantitative,and reliable.An average annual soil loss in UWMR of 25.68,21.04,and 16.80 t ha-1a-1was estimated for 1990,2000 and 2010,respectively,corroborated by comparing spatial and temporal variation in sediment yield.Between 2000 and 2010,a 1.38% average annual increase was observed in the area of lands that lost less than 5 t ha-1a-1,while during 1990-2000 such lands only increased on average by 0.46%.Areas that classified as severe,very severe and extremely severe accounted for 5.68% of the total UWMR in 2010,and primarily occurred in dry areas or grasslands of sloping fields.The reason for the change in rate of soil loss is explained by an increased appreciation of soil conservation by developers and planners.Moreover,we recommend that UWMR watershed adopt further conservation measures such as terraced plowing of dry land,afforestation,or grassland enclosures as part of a concerted effort to reduce on-going soil erosion. The Revised Universal Soil Loss Equation (RUSLE) was applied to assess the spatial distribution and dynamic properties of soil loss with geographic information system (GIS) and remote sensing (RS) technologies. To improve the accuracy of soil-erosion estimates, a new C -factor estimation model was developed based on land cover and time series normalized difference vegetation index (NDVI) datasets. The new C-factor was then applied in the RUSLE to integrate rainfall, soil, vegetation, and topography data of different periods, and thus monitor the distribution of soil erosion patterns and their dynamics during a 30-year period of the upstream watershed of Miyun Reservoir (UWMR), China. The results showed that the new C-factor estimation method, which considers land cover status and dynamics, and explicitly incorporates within-land cover variability, was more rational, quantitative, and reliable. An average annual soil loss in UWMR of 25.68, 21.04, and 16.80 t ha-1a-1 is estimated for 1990, 2000 and 2010, respect ively, corroborated by comparing spatial and temporal variation in sediment yield .etween 2000 and 2010, a 1.38% average annual increase was observed in the area of ​​lands that lost less than 5 t ha-1a-1 while while 1990-2000 such lands only increased on average by 0.46% .Areas that classified as severe, very severe and extremely severe accounted for 5.68% of the total UWMR in 2010, and occurred in dry areas or grasslands of sloping fields. reason for the change in rate of soil loss is explained by an increased appreciation of soil conservation by developers and planners. Moreover, we recommend that UWMR watershed adopt further conservation measures such as terraced plowing of dry land, afforestation, or grassland enclosures as part of a concerted effort to reduce on- going soil erosion.