Nonpoint source (NPS) pollution continues to be the leading cause of US surface water degradation, especially in agricultural areas. In humid regions where variable source area (VSA) hydrology dominates storm runoff, NPS pollution is generated where VSAs coincide with polluting activities. Mapping storm runoff generating areas could allow for more precise and informed targeting of NPS pollution mitigation practices in agricultural landscapes. Previous efforts to do this have relied on simulation models or potentially over-simplifying assumptions about the interactions between rainfall and landscape features that generate storm runoff. Here we used direct measures to empirically derive relationships between topographic wetness indices (TWI) and soil volumetric water content (VWC) and rainfall frequencies to develop runoff risk maps. We surveyed VWC across five agricultural fields in central New York over two years (2012-2014) to develop runoff probability maps based on a soil topographic index (STI). We assumed that the threshold for runoff occurred when the combination of antecedent soil water and rainfall were sufficient to saturate the soil. The linear relationship between VWC and STI was strong for all seasons sampled (spring, summer, autumn). All sites followed a logistic relationship between probability of runoff and STI, although the relative risks between sites shifted from season to season. This work suggests that by developing and using runoff risk maps, the risk of NPS pollution in runoff can be reduced by 70-80% by taking 10% of the agricultural land out of production or halting polluting activities in high risk areas. Reducing the risk of polluted runoff from VSAs depends on the management decisions made at each field, whether the focus is on removing a consistent amount of land from production or setting a threshold for acceptable runoff probability. This analysis can be used to determine the optimal placement of conservation easements or management practices for the protection of water quality.