Within- and cross-center test-retest studies were conducted to study the reliability of a new, semistructured, comprehensive, polydiagnostic psychiatric interview being used in a multisite genetic linkage study of alcoholism. Findings from both studies indicated that reliability for the Semi-Structured Assessment for the Genetics of Alcoholism (SSAGA) was high for DSM-III-R substance dependence disorders, but less so for substance abuse disorders. Reliability of depression was good in both studies, but mixed for antisocial personality disorder (ASP). Findings are presented in terms of specific substance dependence and abuse diagnoses, as well as for depression and ASP. Criterion-specific reliabilities are examined by type of substance used. Although SSAGA was designed to provide for broad phenotyping of alcoholism, review of its new features suggests its suitability for a variety of family studies, not just those focusing on substance abuse.