Abstract: Abstract: In order to make clear the slope soil erosion difference under different rainfall patterns in rocky mountain area of North China, a method combining meteorology and hydrology data with WEPP model to simulate rainfall erosion process was adopted, based on the observation data of 105 individual rainfall events from 2013 to 2015 in field runoff plots of Puwa, Fangshan District, Beijing City. Taking rainfall amount, rainfall duration and average rainfall intensity as the characteristic indices, with R software, 105 natural rainfalls were divided into 3 rainfall patterns A, B and C, and the cluster analysis method was used. The results showed that the erosion ability of the 3 rainfall patterns followed the order of Type B (low frequency, short duration, medium amount of rainfall and strong rainfall intensity) > Type A (medium frequency, long duration, great amount of rainfall and medium rainfall intensity) > Type C (high frequency, medium duration, minor amount of rainfall and low rainfall intensity). However, in the erosion rainfall events, the main rainfall pattern was Type A, which made the highest cumulative contribution rate to soil erosion in the study area, and Type C could hardly cause soil erosion. By analyzing the influence of single rainfall index (rainfall amount, rainfall duration and maximum 30-minute rainfall intensity) on soil erosion, it was found that the erosion ability of 3 types of rainfall in unit time was Type B > Type A > Type C. The critical rainfall amount of soil erosion on brown soil slope in rocky mountain area of North China was 8 mm and the critical rainfall intensity was 9.5 mm/h. In order to explore the influence of time sequences on soil erosion, with WEPP model, the 3 types of rainfall patterns mentioned above were further classified into 4 sub-rainfall types by setting different occurrence time of maximum 30-minute rainfall intensity in the process of rainfall with the WEPP model. Namely, when the value was set to 0, the sub-rainfall type was uniform type. Similarly, the value of 10% corresponded to decreasing type, 50% corresponded to peak type and 90% corresponded to incremental type. Before the WEPP model was used, the soil parameters in the model including effective hydraulic conductivity, critical shear force and erodibility of rill soil were modified on the basis of the value calculated by formula in the model on the physical and chemical properties of the soil. The analysis showed that the time sequences of rainfall intensity had a great influence on the degree of soil erosion. Moreover, under 4 sub-rainfall types, the total soil loss amount was in the order of incremental type > peak type > decreasing type > uniform type. Among them, the difference of soil erosion under sub-rainfall type of Type A was the most significant. In conclusion, this study provides the reference for the establishment of soil erosion prediction models and soil erosion control in brown soil slopes in rocky mountain areas of North China.