Aims: To evaluate the proposed revisions to the DSM-IV alcohol use disorder criteria using epidemiological data. Design, setting and participants: Data came from the 1997 Australian National Survey of Mental Health and Well-Being. The sample consisted of 10641 participants aged 18 years and over. Measurement: Alcohol use disorders were assessed using a revised version of the CIDI 2.0. Alcohol use disorders were assessed in all respondents who indicated that they had used alcohol more than twelve times in the previous twelve months (n=7746). Findings: The proposed introduction of a single alcohol use disorder was supported by CFA. DSM-5 criteria were all indicators of a single underlying disorder. Under DSM-5, the prevalence of alcohol use disorders would increase by 61.7% when compared with those diagnosed under DSM-IV. When investigating the most appropriate diagnostic threshold, the 3+ threshold maximised agreement between DSM-IV and DSM-5 diagnoses, and produced similar prevalence estimates to those yielded by DSM-IV. IRT analyses supported the removal of the Legal criterion whilst provided equivocal results for the Craving criterion. Conclusions: Under the proposed DSM-IV revisions for alcohol use disorders, estimates of the prevalence in the general population would increase substantially. Whereas evidence supports some of the revisions such as a single underlying disorder, others such as the 2+ threshold for diagnosis of alcohol use disorder and the inclusion of a 'craving' criterion may be problematic.