In 1994, DSM-IV will be published, with new criteria for alcohol abuse and dependence. Implications of the changes in criteria for alcohol use disorders were investigated by comparing the diagnoses made by the DSM-IV criteria and DSM-III-R criteria. The study was conducted in a sample of 424 patients in an inpatient alcohol rehabilitation unit in the New York metropolitan area. DSM-III-R and DSM-IV criteria showed similar results and high agreement for any alcohol use disorder (abuse and dependence combined). Alcohol dependence was also consistently diagnosed with DSM-III-R and DSM-IV criteria. Agreement between DSM-III-R and DSM-IV abuse diagnoses was very low. Compared with DSM-III-R, DSM-IV classified over three times as many patients as alcohol abusers, although those with alcohol dependence still overwhelmingly dominated the sample. With some fluctuations, the results were stable over Black, Hispanic, and White subgroups of patients. The direction of the findings was consistent with results from a national general population survey in that the prevalence of alcohol abuse increased in both studies. However, the clinical results alone would not have suggested the marked changes in relative prevalence of abuse and dependence that occurred in the general population when DSM-IV criteria were used in place of DSM-III-R. Research on diagnostic criteria limited to patient samples omits important information on the implications of changing aspects of the diagnostic criteria. The need for a coherent theory of alcohol abuse is highlighted.