In the context of climate change, modeling the spatiotemporal evolution of soil erosion and examining its responsiveness to climatic factors were essential for addressing climate change and improving disaster prevention and mitigation efforts. Existing studies primarily focused on the impacts of climate variability, slope, and vegetation restoration on soil erosion in the Loess Plateau. However, few studies simultaneously considered the interactions among these driving factors and their direct and indirect impacts on soil erosion. This study used data from meteorological stations, land use/land cover, and soil texture to analyze the spatiotemporal characteristics of climatic factors using Theil－Sen Median trend analysis and Mann－Kendall tests. The InVEST (Integrated Valuation of Ecosystem Services and Tradeoffs) model was applied to simulate the spatial and temporal distribution of soil erosion in the Loess Plateau for the years 1990, 2000, 2010, and 2020. Additionally, the optimal parameters-based geographical detector model and the partial least squares structural equation modeling were used to assess the intensity and pathways through which climatic factors influence soil erosion, while considering natural and vegetation factors. The results showed that climatic factors exhibited both temporal and spatial heterogeneity: precipitation decreased at a rate of ?55.96 mm per decade from 1990 to 2000 but increased at a rate of 55.99 mm per decade from 2000 to 2020. During the study period, annual precipitation, precipitation intensity index, number of heavy rainfall days, extreme precipitation events, mean temperature, and minimum temperature increased at rates of 26.15 mm per decade, 0.26 mm per day per decade, 0.56 days per decade, 15.21 mm per decade, 0.32?°C per decade, and 0.40?°C per decade, respectively. Spatially, precipitation declined in 86.36% of the study area between 1990 and 2000, whereas it increased in 97.42% of the region between 2000 and 2020. From 2000 to 2020, extreme precipitation indicators generally rose across the entire study area. Temperature increases were most prominent in the eastern and western regions, reflecting a clear trend of warming and moistening, accompanied by intensified precipitation extremes. From 1990 to 2020, the soil erosion modulus in the Loess Plateau initially first declined and then increased, with soil erosion reaching 219 million tons in 2020. The optimal parameters-based geographical detector analysis revealed that slope, precipitation, and vegetation cover are the primary drivers of soil erosion, with the explanatory power of precipitation increasing from 0.11 in 1990 to 0.18 in 2020. The partial least squares structural equation modeling analysis further indicated that temperature primarily influenced soil erosion indirectly by affecting precipitation. Both precipitation and natural factors had direct positive contributions to soil erosion, whereas vegetation factors had a direct negative impact, although this effect decreased by 0.02 from 2010 to 2020. These findings highlight the significant impact of warming-wetting trends and the intensification of extreme precipitation events on soil erosion. Therefore, future soil erosion prevention and sustainable development efforts should integrate climate adaptation strategies with regional development plans to effectively address the challenges posed by ongoing climate change.