Surplus nitrogen (N) estimates, principal component analysis (PCA), and end-member mixing analysis (EMMA) were used in a multisite comparison contrasting the fate of N in diverse agricultural watersheds. We applied PCA-EMMA in 10 watersheds located in Indiana, Iowa, Maryland, Nebraska, Mississippi, and Washington ranging in size from 5 to 1254 km with four nested watersheds. Watershed Surplus N was determined by subtracting estimates of crop uptake and volatilization from estimates of N input from atmospheric deposition, plant fixation, fertilizer, and manure for the period from 1987 to 2004. Watershed average Surplus N ranged from 11 to 52 kg N ha and from 9 to 32% of N input. Solute concentrations in streams, overland runoff, tile drainage, groundwater (GW), streambeds, and the unsaturated zone were used in the PCA-EMMA procedure to identify independent components contributing to observed stream concentration variability and the end-members contributing to streamflow and NO load. End-members included dilute runoff, agricultural runoff, benthic-processing, tile drainage, and oxic and anoxic GW. Surplus N was larger in watersheds with more permeable soils (Washington, Nebraska, and Maryland) that allowed greater infiltration, and oxic GW was the primary source of NO load. Subsurface transport of NO in these watersheds resulted in some removal of Surplus N by denitrification. In less permeable watersheds (Iowa, Indiana, and Mississippi), NO was rapidly transported to the stream by tile drainage and runoff with little removal. Evidence of streambed removal of NO by benthic diatoms was observed in the larger watersheds.