Soil erosion in croplands poses a strong threat to food security and environmental sustainability. A comprehensive understanding of the driving forces and their contributions to changes in cropland soil erosion provides solutions for effective mitigation strategies. The factors of embodied soil erosion changes have rarely been systematically described on the regional scale. We combined the Modified Universal Soil Loss Equation (RUSLE) with the multi-regional input-output (MRIO) model in the first step to analyze the flow of embodied cropland soil erosion in the trade network. The structural decomposition analysis (SDA) model was used to decompose the embodied cropland soil erosion into six driving factors: natural factor, human factor, and socio-economic activities. The SDA results reflect that the nature factor (8.0 % and 6.0 % in two periods from 2012 to 2017) and the final demand factor (17.3 % and 9.9 % in two periods from 2012 to 2017) facilitated the national cropland soil erosion increasing. The human factor and the economic production structure factor always reduced the growth of cropland soil erosion. This study contributes to understanding erosion drivers at provincial and national levels in China, and guides the policy interventions for sustainable soil and water management. © 2024 Elsevier Ltd