This study tests the hypothesis that lakes in watersheds dominated by row-crop agriculture (e.g., maize or soybeans) have systematically higher N : P than lakes in watersheds with large tracts of pasturelands. Current biogeochemical models of eutrophication suggest that agricultural nitrogen and phosphorus fluxes lead to a systematic decline in the N : P of receiving waters. In contrast, different agricultural activities (i.e., row-cropping vs. animal agriculture) use greatly divergent N and P amendments, and fluxes from agricultural watersheds diverge through a broad range of observed N : P (i.e., sub-Redfield to >100). Animal agriculture leads to low N : P fluxes and row-cropping to high N : P. The connection between agricultural watershed land use and lake nutrients stoichiometry was tested in a highly agricultural region of the United States (Iowa) on 113 lakes in watersheds with different amounts of row-crop (0%-95%) and pastureland (0%-36%). Multiple regression analysis shows that lakes in watersheds with large areas in pasturelands have low N : P, whereas lakes in watersheds dominated by row-cropping have systematically high N : P. Lakes in watersheds with >30% pasture had the lowest N : P, approaching Redfield levels. N : P was most frequently high (>50 as atoms) in lakes with >90% of their watersheds in row-crop agriculture. The dynamics of agricultural practice necessitates the inclusion of real-world differences among agricultural systems in nutrient stoichiometric models. Intensive row-crop agriculture yields N : P stoichiometry at high levels usually observed in pristine headwaters and open oceans, whereas increased animal agriculture will drive N : P to low levels usually associated with cyanobacterial blooms.