In recent years, as control of industrial point source pollution has gradually increased, agricultural nonpoint source pollution (ANPSP) has become an important focus of water pollution control. ANPSP is closely related to challenges in the urbanization process, though urbanization has mitigated water pollution control to a certain extent. In this study, taking the Taihu Lake Basin of China as an example, the dynamic interrelation between urbanization and ANPSP is macroscopically displayed by the Vector autoregressive (VAR) model, and the coevolution characteristics of the two are quantitatively measured by the decoupling index. Then, the Logarithmic Mean Divisia Index (LMDI) decomposition method is used to explore the influence of urbanization's accompanying effects, including pollution intensity, economic structure, output level and population factors, on ANPSP. The results showed that (1) since 2006, total nitrogen (TN) has continued to decline, while total phosphorus (TP) initially increased then decreased. The main pollution sources were livestock and poultry breeding and planting, which showed a trend of alternating growth and decline. At the same time, the contributions of rural life and aquaculture to pollution gradually increased. (2) Urbanization has a positive effect on ANPSP control after reaching a certain stage, but there is a possibility of pollution rebound; urbanization mitigates ANPSP, but still contributes to environmental pollution. (3) The decoupling between urbanization and ANPSP is achieved across the basin. However, the decoupling level between TP and urbanization lagged behind the decoupling level between TN and urbanization. Meanwhile, the spatiotemporal heterogeneity of the TP decoupling index is more evident than that of the TN index. (4) The intensity effect and structure effect promote the strong decoupling between urbanization and ANPSP, while the output effect and population effect make the decoupling relationship evolve to adverse states such as expansive negative decoupling and expansive coupling.