Accelerated eutrophication and nutrient loads in the lakes are of major concern for human health and environment. This study was undertaken for modeling the non-point source pollution of Taihu lake basin in eastern China. The SWAT model having an interface in ArcView GIS was employed. Model sensitive parameters related to hydrology and water quality were obtained by sensitivity analysis, and then calibrated and validated by comparing model predictions with field data. The GIS showed good potential for parameterization of hill-slopes, channels, and representative slope profiles for SWAT model simulations. in a monthly and daily time step, the model's Nash-Sutcliffe coefficient (E) and the coefficient of determination (R-2) indicated that values of simulated runoff, NH4+-N and total phosphorus were acceptably closer to the measured data. Surface water parameters especially CN, Soil-AWC and ESCO were the most sensitive and had more recognition in the model. It is concluded that runoff carrying N and P nutrients from chemical fertilizer inputs in agricultural areas is the major contributor to NPSP in the lake basin. So. decrease in excessive use of N and P fertilizers and their synergism with organic manures is recommended that would significantly reduce nutrient pollution in the take ecosystem. (C) 2009 Elsevier B.V. All rights reserved.