This paper presents a stochastic framework for the assessment of groundwater pollutants potential of nonpoint source pesticides. A conceptual relationship is presented that relates seasonally averaged groundwater recharge to soil properties and depths to the water table. The analytical relationship shows a linear association with the soil saturated hydraulic conductivity, and predicts less recharge for shallower water tables. The stochastic framework utilizes first-order approximations of the mean and variance of each of the recharge and the residual mass emissions in the soil. Soil and chemical properties and related environmental factors, which affect the fate and transport of pesticides, and the depth to the water table are modeled as random variables. The environment-fate models are integrated with a GIS, and the stochastic framework is applied to assess potential nonpoint-source vulnerability of shallow groundwater to the pesticide dicamba in Mid-Atlantic coastal plain agricultural watersheds. It is shown that recharge estimates on the basis of the SCS abstraction method resulted in lower expected dicamba concentrations in groundwater, than when leaching is based on the conceptual model. In the analysis, the first-order estimates of the variance of dicamba mass emissions and estimated groundwater concentrations, were of order of magnitude of their respective means and greater. While some of the mean groundwater concentrations showed significant residual levels of dicamba, these values however should be viewed with greater uncertainty. Given the expected uncertainties in the input data and model errors, regulatory decisions and environmental land-use planning should take into account estimates of the uncertainties (variances) associated with predictions of groundwater vulnerabilities.