Abstract: Quantifying the mechanisms driving the runoff-sediment relationship under different rainfall intensities is crucial for overcoming cumulative effect limitations and accurately characterizing the evolution of watershed erosion dynamics. This study focuses on the Beiyu River, a tributary within the Jialing River system. Utilizing meteorological, hydrological, and land surface data from 2007 to 2020, 264 erosive rainfall events were identified, and a partial least squares-structural equation model (PLS-SEM) was constructed to quantitatively analyze the variation characteristics and dominant drivers of the runoff-sediment relationship across different rainfall intensities. The results indicated that the Beiyu River watershed exhibits a typical low runoff coefficient and high sediment concentration. High-intensity events (daily maximum precipitation > 30 mm) were the critical drivers of sediment transport, contributing to approximately 44% of the average annual total sediment load on a per-event basis. As precipitation intensity increased, the water and sediment relationship transitioned from a highly dispersed state (R² = 0.07—0.09) to a significant power function relationship (R² = 0.94). PLS-SEM analysis revealed that the explanatory power of runoff for sediment variation increased significantly with precipitation intensity (the path coefficient rose from 0.519 to 0.922), with sedient transport primarily regulated by the synergy of antecedent precipitation and runoff characteristics. In addition, the regulatory effects of land surface factors, including temperature, vegetation, and sediment connectivity, became more prominent in determining suspended sediment concentration during high-intensity rainfall events. These findings provide a scientific basis for the precise prevention and control of soil erosion in small mountainous watersheds.