Abstract: Abstract: Agricultural non-point source pollution is one of main ecological problems in the Three Gorge Reservoir Region (TGRE), and greatly impact ecological security and socio-economic development of TGRE. In this paper, main stream of the Yangtze River was supposed to be as ultimate collection area of agricultural non-point source pollutants. As an important source landscape, cultivated land including paddy field and dry land was classified into 4 grades by using buffer tool of ArcGIS to build buffers in the TGRE at the distance of 0-20, 20-40, 40-60 and 60-80 km from main stream of the Yangtze River. Considering the resistance function of different influence factors, the resistance base surface impacting the agricultural non-point source pollution was constructed based on the selection of main natural influence factors, including topography (relative elevation, relative slope and surface roughness), meteorology (rainfall erosion), hydrology (topographic wetness index), soil (soil erosion) and vegetation (vegetation coverage). And then the minimal cumulative resistance (MCR) model was applied to obtain the resistance surface of source landscape of different grades, by which source-sink risk patterns were identified. In the end, according to the resistance surface, source-sink risk pattern was classified into 5 grades (extremely low risk, low risk, medium risk, high risk and extremely high risk) by the classification of natural breaks law to analyze their spatial characteristics. The results showed that: 1) In the TGRE, source cultivated land of the first level, which was located in the buffer at the distance of 0-20 km from main stream of the Yangtze River, occupied more than 50% of the total cultivated land area, and the higher grade of source cultivated land corresponded to the smaller space distribution; the distribution of first-level source cultivated land in Chongqing reservoir area was more than that in Hubei reservoir area, and the distribution of dry land was more than paddy field; 2) Surface resistance changes of different-grade sources were mainly influenced by spatial distance; the value of resistance surface was small in the buffers located at the source landscape, and became larger from source landscape to the outside, and the resistance surface value of paddy field was larger than that of dry land due to the impact of spatial distance on resistance surface; 3) The MCR model was applied to classify source-sink risk pattern into 5 grades, including extremely high risk zone (21 706.13 km2 in TGRE, similarly hereinafter), high risk zone (7 464.65 km2), medium risk zone (16 257.75 km2), low risk zone (2 221.61 km2) and extremely low risk zone (10 311.6 km2), which indicated that there was a trend of high risk in the risk pattern of source sink by risk zone area statistics of districts and counties; high risk areas were mainly distributed in the area of parallel ridge valley, while low risk areas were mainly located in Qinba mountain area and Wuling mountain area with the remote distance from the Yangtze River. The results are helpful to evaluate the risk degree and rank of non-point source pollution produced by cultivated land from the angle of resistance surface impacting non-point source pollution, and can provide the policy-making basis for preventing and controlling agriculture non-point source pollution scientifically.