The Erosion Productivity Impact Calculator (EPIC) model was tested using four years of field data collected at a site near Lamberton, Minnesota, under three different crop rotations: continuous corn (Zea mays L.) or CC, soybean (Glycine max L.)-corn (SC), continuous alfalfa (Medicago sativa L.) or CA. The model was evaluated by comparing measured versus predicted subsurface drainage flow (tile flow), nitrate-nitrogen (NO3-N) loss in tile flow, residual NO3-N in the soil profile, crop N uptake, and yield. Initially, EPIC was run using standard Soil Conservation Service (SCS) runoff curve numbers (CN2) suggested for the soil type at the site. Two different SC runs were performed with a nitrogen fixation parameter denoted as parm(7) set at either 1.0 or 0.3, reflecting uncertainty for this parameter. Under this scenario, EPIC accurately tracked monthly CC and SC variations of tile flow (r2 = 0.86 and 0.90) and NO3- N loss (r2 = 0.69 and 0.52 or 0.62). However, average annual CC and SC tile flows were under-predicted by 32 and 34 percent, and corresponding annual NO3-N losses were under-predicted by 11 and 41 or 52 percent. Predicted average annual tile flows and NO3-N losses improved following calibration of the CN2; CC and SC tile flow under-predictions were -9 and -12 percent while NO3-N losses were 0.6 and -54 or -24 percent. In general, EPIC reliably replicated the impacts of different crop management systems on nitrogen fate; e.g., greater N loss under CC and SC than CA, and less residual soil N under CA as compared to the other cropping systems. The simulated CA monthly tile flows and NO3-N losses also compared poorly with observed values (r2 values of 0.27 and 0.19). However, the predicted CA annual drainage volumes and N losses were of similar magnitude to those measured, which is of primary interest when applying models such as EPIC on a regional scale.