A SWAT modeling framework has been constructed for the Boone River Watershed (BRW) in north central Iowa, to support further testing of SWAT and analyses of alternative management practice and/or cropping system scenarios. The BRW covers over 237,000 ha and is an intensively cropped region dominated by corn and soybean production. Nitrate losses are of particular concern in the BRW, especially through subsurface tiles that drain the predominantly flat landscapes that persist throughout the watershed. The modeling system features an intensive set of management, land use, and soil data developed at the Common Land Unit (CLU) level. The SWAT model was tested using two different hydrologic simulation approaches for the BRW, that were based on the standard runoff curve number (RCN) option versus a new alternative RCN option available in version 2005 of SWAT. These two different approaches were used to reflect differing assumptions regarding the relative contributions of surface runoff and baseflow to the total BRW streamflow. Strong annual and monthly R2 and Nash-Sutcliffe modeling efficiency (E) statistics were found for both the 1986-1996 calibration and 1996-2006 validation periods, which ranged from 0.74 to 0.99. The R2 and E statistics determined for the calibrated annual and monthly sediment, nitrate, organic nitrogen, and total phosphorus loads for the period of 2000-2006 were also generally strong for the SWAT simulations that were performed with the standard RCN approach, ranging from 0.50 to 0.92 with most exceeding 0.70. However, the accuracy of the predicted pollutant loads generally declined when the alternative RCN approach was used, especially for the organic nitrogen estimates. The results show that specific calibration is necessary for pollutant-related input parameters for the alternative RCN approach, in order to obtain improved results. The results also show weaknesses in the overall nitrogen balance predicted for the SWAT simulations, especially for the approach based on the standard RCN method. Comparisons with historical crop yields revealed that the model underpredicted corn yields, especially for the standard RCN approach, and that there is a need to update crop parameters in SWAT to more accurately simulate current corn and soybean yields in the region.