BACKGROUND The validity of self-report data is an area of continuing concern in the substance abuse treatment field. It is uncertain how well self-report of alcohol relapse corresponds with more objective indices. METHODS We compared the self-report of alcohol relapse to collateral reports and biological indices of relapse. Twelve-month post-treatment follow-up data were collected from 94 male and female alcohol dependent veterans and 93 of their respective collateral contacts. Biological indices included breathalyzer data, and the blood enzymes aspartate aminotransferase, y-glutamyltransferase, and alanine aminotransferase. RESULTS A collapsed factor of the more objective indices of use was moderately associated with self-report. A logistic regression analysis revealed that only collateral reports of use predicted the self-report of alcohol relapse. The specificity of collateral report was 82.4% and the sensitivity was 71.9%. CONCLUSIONS Collateral informants serve an important function in supporting the validity of self-report of abstinence versus relapse.