Soil erosion poses a threat to food security and ecosystem services， and is one of the severe environmental problems facing China. It is affected by both natural factors and human activities. At present， a large number of studies in China have focused on the driving effect of socio-economic factors on soil erosion， but there are still insufficient in studying the spatial non-stationary relationship between the two and paying attention to the scale differences of affecting factors. 346 prefecture-level cities in China were used to study the complex driving mechanisms of socio-economic activities on soil erosion. The soil erosion prediction （RUSLE） model and multiscale geographically weighted regression （MGWR） model were used to investigate the spatial heterogeneity of soil erosion in various cities in China using 2017 as a reference year. The spatial driving effects of socio-economic factors on rates of soil erosion in prefecture-level cities across China and the differences in the scale of action between factors were studied. Results showed that the spatial distribution of the rate of soil erosion in various prefecture-level cities in China had a significant positive spatial correlation. Hotspots of soil erosion were mainly distributed in the west， northeast， the Yunnan-Guizhou Plateau， Sichuan Basin and the Loess Plateau. Compared with models based on global regression and traditional geographically weighted regression models， MGWR significantly improved the explanatory power of socio-economic variables on the rate of soil erosion， with a model fitting goodness of 0.87. From the perspective of driving factors， the direction of the effect of each driving factor on the rate of soil erosion in various prefecture-level cities in China had structural differences with changes of spatial location except for the factor of regional gross domestic product per capita. On average， population density was the factor contributing the most to the rate of soil erosion in various prefecture-level cities in China. The rate of soil erosion in prefecture level cities in China was more susceptible to the influence of multiple cropping indices in the western region， while the driving mechanism of socio-economic factors on the rate of soil erosion in the eastern region was more complex， and the spatial scale differences of different driving factors were more obvious. It is indicated that decision makers should fully consider the spatial heterogeneity of the impact of human activities on soil erosion to achieve the sustainable development of soil and water conservation.