Computer simulation models are regarded as important tools to address concerns about the environmental impact of agricultural activities and have increasingly been used to complement field research. The CERES-Maize and CROPGRO-Soybean, two of the main crop growth models incorporated into the Decision Support System for Agrotechnology Transfer (DSSAT), were developed to simulate the effects of environmental conditions, different management strategies, and genetics on crop growth and development. These models currently do not have an option for calculating erosion and phosphorus (P) dynamics. In this study, the Modified Universal Soil Loss Equation (MUSLE) and sediment-bound P model were incorporated into CERES-Maize and CROPGRO-Soybean models. The evapotranspiration component of the CERES model was also evaluated to test the ability of the model to simulate the effects of shelterbelts on corn transpiration. Two years of sap flow data collected from corn plants in a sheltered (3H) and unsheltered (25H) areas in a field near Ogden, Iowa, were used for model testing. Results showed that the CERES-Maize model simulated daily plant transpiration well in both sheltered and unsheltered areas with RMSE of 0.53mm d-1 (3H) and 0.75mm d-1 (25H) in 2001, and 0.85mm d-1 (3H) and 0.59mm d-1 (25H) in 2002. Average coefficient of determination (r2) for 2001 and 2002 was 0.90 and 0.73, respectively. After erosion and sediment-bound P subroutines were incorporated into CERES-Maize and CROPGRO-Soybean, the models were calibrated and tested using five years of data collected from two field sites in a watershed in Tama County, Iowa. Results showed that generally both models tended to over or underpredict daily sediment and sediment-bound P losses from fields, but they simulated seasonal values reasonably well. For CERES model, the error between simulated and measured seasonal sediment losses was less than 10% in three out of the five years, while in four out of the five years, the difference between simulated and measured sediment loss was less than 15% for the CROPGRO. The study indicated that both models did not seem to give good estimates of daily sediment and P losses, but they can be used to simulate the long-term (seasonal) losses with reasonable accuracy.