Neonicotinoids (NEOs) have been widely used for decades but concurrently pose substantial risks to nontarget organisms. Identifying the key pathways of NEOs is crucial for mitigating risks, whereas most multimedia models exhibit limitations in quantifying the fate of chemicals at large scales and over long periods. Here, we developed a novel multimedia model that incorporated stratified agricultural soil compartments, historical accumulation, and an enhanced hydrological module. The new model was applied to 45 tertiary watersheds in the Yangtze River Basin (YRB), focusing on imidacloprid (IMI) as a representative neonicotinoid and spanning a decade (2013-2022). Highest concentrations in agricultural soils, surface water, and sediment pore water were 34.1 ng g- 1, 53.5 ng L-1, and 1540 ng L- 1, respectively. IMI inputs into the environment were predominantly from paddy and dryland fields, totaling 192 tonnes, with rice and wheat as the main sources. Legacy IMI accounted for 26.3 %, 9.58 %, and 13.3 % of the inputs to agricultural soils, surface water, and sediment, respectively. Approximately 28.9 tonnes of IMI entered surface water via runoff, with 7.86 tonnes (2.44 % of the total input) discharged into the East China Sea. Ecological risks were at moderate to high levels and were expected to intensify with continued input and historical accumulation. This model performed well in terms of identifying agricultural sources, quantifying legacy compounds, and enhancing interwatershed hydrological connectivity, which can provide valuable insights for NEO risk management in complex agricultural watersheds.