Synthesis and analysis of system level processes and their interactions are critical in ensuring environmental and economic sustainability in agriculture. Agricultural system models help synthesize knowledge from component experiments of the system, and help predict its behavior in response to management. In this study we used the RZWQM and RZWQM-DSSAT hybrid models for modeling the long-term (1978 to 2003) impacts of four tillages, three crop rotations, and eight nutrient management practices on crop production, and soil and tile water quality. These experiments were conducted on 36, 0.4 ha plots on the Clyde-Kenyon-Floyd soil association in northeastern Iowa. Data from 1990 to 2003 from one of the plots (No. 25) with continuous corn (Zea mays L.) from 1978 to 1992 and soybean (Glycine max L. Merr.) - corn rotations from 1993 to 2003, constantly in the no-till system was used for model calibrations. Both RZWQM and RZWQM-DSSAT hybrid models have the same soil water and nutrient simulation modules hence were calibrated only once and used for running both the models. When calibration accuracies of soil water in different soil layers, water table, daily tile flow, NO3-N in tile water, and NO3-N in soil for both models were comparable, grain yield, biomass, and grain and plant N uptakes predictions of RZWQM-DSSAT hybrid model was more accurate. Accuracies of grain yield and biomass predictions of both the models deteriorated slightly in the 35 validation plots. Better measured grain yield due to Late Spring Nitrogen Test (LSNT) over pre-plant N application was simulated well by both the models but RZWQM predictions were closer to measured. Measured impacts of various tillage, nutrient and crop rotation management practices on grain yield, biomass, tile N, residual soil N, water table, tile flow, and soil water were predicted with reasonable accuracy by both the models with some exceptions.