Groundwater is an important source of freshwater supply in Thailand, particularly in dry season for non-irrigated area. Unfortunately, our groundwater supply has generally not been adequately protected from natural and anthropogenic contaminants. Groundwater contamination is a complex process and full of uncertainty at regional scale. Assessments of the vulnerability of groundwater to contamination range in scope and complexity from simple, qualitative, and relatively inexpensive approaches to rigorous, quantitative, and costly assessments. This paper presents a stochastic framework for risk assessment of agricultural contamination in soil and groundwater using GIS-based geostatistical models accounted for aquifer heterogeneity and temporally averaged rainfall-runoff data. A conceptual relationship was employed to relate seasonally averaged groundwater recharge to soil properties and depths to the water table. The stochastic model utilizes first-order approximations of mean and variance of groundwater recharge to the aquifer underneath, while depth to the water table is modeled as random variables. The model is integrated with a GIS and the stochastic framework is applied to assess the groundwater vulnerability (i.e., risk of shallow groundwater to contamination) based on DRASTIC deterministic model concept. Development of an integrative vulnerability assessment methodology can be useful to effectively manage (including prioritization of limited resource allocation to monitor high risk areas) and protect this valuable freshwater source. Optimization of stochastic groundwater vulnerability to some objective functions is also undertaken using cross validation technique with root mean square error (RMSE) and determination coefficient (R2) for effectiveness evaluation.