Agrichemicals (herbicides and nitrate) are significant sources of diffuse pollution to groundwater. Indirect methods are needed to assess the potential for groundwater contamination by diffuse sources because groundwater monitoring is too costly to adequately define the geographic extent of contamination at a regional or national scale. This paper presents examples of the application of statistical, overlay and index, and process-based modeling methods for groundwater vulnerability assessments to a variety of data from the Midwest U.S. The principles for vulnerability assessment include both intrinsic (pedologic, climatologic, and hydrogeologic factors) and specific (contaminant and other anthropogenic factors) vulnerability of a location. Statistical methods use the frequency of contaminant occurrence, contaminant concentration, or contamination probability as a response variable. Statistical assessments are useful for defining the relations among explanatory and response variables whether they define intrinsic or specific vulnerability. Multivariate statistical analyses are useful for ranking variables critical to estimating water quality responses of interest. Overlay and index methods involve intersecting maps of intrinsic and specific vulnerability properties and indexing the variables by applying appropriate weights. Deterministic models use process-based equations to simulate contaminant transport and are distinguished from the other methods in their potential to predict contaminant transport in both space and time. An example of a one-dimensional leaching model linked to a geographic information system (GIS) to define a regional metamodel for contamination in the Midwest is included.