Quantifying the pollutant loads in rainfall-runoff plays a critical role in non-point water pollution control at the watershed level. In this study, rainfall-runoff experiments were conducted on various land use types in the western bank area of Wangyu River to localize the initial abstraction ratio in the SCS-CN model. Then, by incorporating the remote sensing interpretation of land use types and the measurement of pollutant concentration in the rainfall-runoff, this study explored the temporal and spatial patterns of four kinds of pollutants, namely total phosphorus (TP), total nitrogen (TN), ammonia nitrogen (NH₃-N) and chemical oxygen demand (COD). The results indicate that the recommended initial abstraction ratios for grassland/forest/farmland and impervious surface in the study area are 0.3 and 0.9, respectively. Besides, the rainfall-runoff depths and pollutant loads are found to be significantly and positively correlated with the rainfall depths. Along with the decrease of annual rainfall in this region, the annual amount of TP, TN, NH₃-N and COD loads in rainfall-runoff have reduced from 190, 1359, 445 and 16041 t in 2017 to 118, 949, 314 and 11250 t in 2019, respectively. Rural resident land, among all the land use types, has the highest TN and COD loads per area, while grassland and forest rank the bottom. This study provides precise information about pollutants in rainfall-runoff to facilitate the development of water pollution control strategies in Wangyu River watershed and can be used as a methodological reference for future quantification of non-point source pollution practices in plain areas with dense river networks.