Non-point source (NPS) pollution has become the major reason for water quality deterioration. Due to the differences in the generation and transportation mechanisms between urban areas and rural areas, different models are needed in rural and urban places. Since land use has been rapidly changing, it is difficult to define the study area as city or country absolutely and the complex NPS pollution in these urban–rural mixed places are difficult to evaluate using an urban or rural model. To address this issue, a fuzzy system-based approach of modeling complex NPS pollutant is proposed concerning the fuzziness of each land use and the ratio of belonging to an urban or rural place. The characteristic of land use, impact of city center and traffic condition were used to describe spatial membership of belonging to an urban or rural place. According to the spatial membership of belonging to an urban or rural place, the NPS distributions calculated by the urban model and rural model respectively were combined. To validate the method, Donghu Lake, which is undergoing rapid urbanization, was selected as the case study area. The results showed that the urban NPS pollutant load was significantly higher than that of the rural area. The land usage influenced the pollution more than other factors such as slope or precipitation. It also suggested that the impact of the urbanization process on water quality is noteworthy.