Groundwater vulnerability is a central concept in pollution risk assessment, yet its estimation has been largely a matter of expert judgment. This work applies a method for the direct calculation of vulnerability from monitoring well observations of pesticide concentrations. The method has two major advantages: it is independent of the compounds being examined, and it has a direct probabilistic interpretation making it ideal for risk assessment. The methodology was applied to data from a groundwater monitoring program in the midwestern United States. The distribution of the vulnerabilities was skewed toward zero. Spatial distribution of the vulnerabilities shows them to be controlled by both regional and local factors. Methods are presented for estimating the necessary sample sizes for vulnerability studies. The further application of the approach developed in this study to understanding groundwater pollution is discussed.