The purpose of this study was to compare DSM-III-R and proposed formulations of DSM-IV diagnoses of alcohol use disorders in a clinical sample with the same assessment methods used in a large national survey previously reported. Issues included the number of symptoms required for abuse and dependence diagnoses, the effects of requiring evidence of physiological dependence to make the dependence diagnosis (Option 1 vs. Option 2), whether to require a duration criterion, and the concept of alcohol abuse. The diagnostic criteria proposed in the 1991 DSM-IV options book were the criteria investigated in this study. Concordance between DSM-III-R and DSM-IV was closest when the form of DSM-IV used was most similar to DSM-III-R. The duration criterion had much less effect on a dependence diagnosis in this clinical sample than in the general population. DSM-IV Option 2 for alcohol dependence excluded a number of cases from the dependence diagnosis who received such a diagnosis under DSM-III-R and DSM-IV Option 1. Abuse continued to function as a residual category, especially noticeable under increasingly restrictive definitions of dependence, and was rare in both the clinical and general population sample. Implications of the findings for research and other purposes are discussed.