Determining the groundwater contribution of nonpoint source pollution at a watershed scale is a challenging issue. In this study, we utilized a top‐down approach to characterize representative groundwater response units (GRUs) based on land use and landscape position (e.g., upland, sideslope, or floodplain) in the 275‐km2 Clear Creek Watershed, Iowa. Groundwater monitoring wells were then established along downslope transects in representative GRUs. This unique combination of top‐down/bottom‐up approaches allowed us to estimate groundwater pollutant loads at the watershed scale with minimal monitoring. For the 2015 study period, results indicated that more groundwater recharge occurred in the floodplain (404 mm) compared to the uplands or sideslopes (281 and 165 mm, respectively), irrespective of land use. Recharge in the floodplains consisted of 37% of the annual precipitation, whereas upland wells averaged 26% and sideslopes averaged 15% of the annual precipitation. Less recharge was found to occur beneath perennial grass compared to row crop and urbanized areas. Baseflow discharge accounted for 69% of the total NO3‐N exported from the Clear Creek Watershed, with row crop areas contributing approximately 95% of the annual load. Orthophosphorus (OP) yields were approximately 0.72 kg/ha beneath urban and suburban areas, three times higher than those in row crop or perennial areas. Urban and suburban areas accounted for 21.4% of groundwater orthophosphorus and chloride loads in the watershed compared to only 8.5% of the land area. Overall, the groundwater load allocation model for baseflow nutrient discharge to Clear Creek can be used to target future nonpoint source load reduction strategies at the watershed scale. The use of GRUs can pinpoint better areas of concern for controlling nutrient loads.