The use of computerized matching of proteins in the analysis of multiple two-dimensional electrophoresis (2DE) gels creates volumes of data that are readily accessible for exploratory analysis. When these data are used in health-effects studies or in studies to identify factors associated with particular diseases, hundreds or even thousands of hypotheses can be tested. Interpreting so many hypothesis tests requires some preliminary statistical evaluations of the data. In addition, prior to the preliminary statistical evaluations and subsequent hypothesis tests, accurate protein quantification and correct protein matching must be verified. In this report we present an approach used at the Centers for Disease Control to address these issues. This approach consists of a randomized experimental design incorporating replicate gels for each specimen, gel image analysis, protein matching, editing, Boolean unions of all gels to obtain correspondences and contradictions of match identification numbers, resolution of correspondences and contradictions, statistical tests to identify outliers, and finally an assessment of statistical and practical significance to focus attention on the proteins most likely to be associated with the effects under study. We illustrate our approach with data from an exploratory exposure-response study.