Abstract: The Qianhe River basin, a main tributary of the Weihe River in the southern Loess Plateau, exhibits highly complex and spatially heterogeneous soil erosion drivers due to influences from natural environments and intensive human activities. To date, targeted quantitative analyses remain scarce. We employed the RUSLE (Revised Universal Soil Loss Equation) model to analyze the spatiotemporal changes in soil erosion in the Qianhe River basin from 2000 to 2020, and used the Geographically Weighted Regression (GWR) model and GeoDetector analysis to identify the key factors on soil erosion and their interactions. Results indicate that soil erosion in the basin showed an overall decreasing trend from 2000 to 2020, during which the soil erosion modulus decreased from 12.46 to 10.58 t·hm⁻²·a⁻¹, being characterized mainly by erosion in slight to light level. The spatial distribution pattern exhibited lower erosion in the periphery and greater erosion in the central area. Regarding the driving mechanisms, among natural factors, topographic elements (slope, elevation) provided fundamental constraints for the regional erosion. Among human activity factors, land use emerged as the primary determinant of soil erosion in the basin with the explanatory power of 0.235. The interaction between slope and land use type exhibited the highest explanatory power at 0.307 for soil erosion. Risk detection identified areas prone to severe soil erosion as those with slopes between 8°～15°, elevations between 1000～1500 m, rainfall erosion potential of 1400～1700 MJ·mm·hm⁻²·h⁻¹·a⁻¹, and grassland land use. This study provided an important data support and theoretical guidance for erosion prevention and control in the watershed.