Aims: To identify an optimal brief version of the Alcohol Use Disorders Identification Test (AUDIT) and to evaluate its effectiveness as a screening test for problem drinking (PD), alcohol use disorders (AUD) and alcohol dependence (AD). Methods: A multicenter trial was conducted from March, 2010, to January, 2012, in 101 normal comparison, 203 risky drinking, 92 alcohol abuse and 101 AD men aged below 65 years of age in a Korean population. For the purposes of screening, risky drinking, alcohol abuse and AD were grouped: all the three grouped as PD and the latter two grouped as AUD. Logistic regression analysis was used to determine the items among the 10-item AUDIT that provided information predictive of PD, AUD and AD. Receiver operating characteristic (ROC) curve analysis was performed to investigate the discrimination ability of the brief versions of AUDIT, 10-item AUDIT and Cut-down, Annoyed, Guilt, Eye-opener as a screen for PD, AUD and AD. Areas under the ROC curve were compared between tests according to the method suggested by Hanley and McNeil. Results: The 5-item AUDIT (AUDIT-5: AUDIT items 2, 4, 5, 9 and 10) was obtained by stepwise multiple regression analyses for each screening. AUDIT-5 exhibited an AUD screening accuracy significantly superior to that of the 10-item AUDIT, but other brief versions of AUDIT and CAGE did not. Furthermore, AUDIT-5 had a high PD and AD screening accuracy equivalent to that of the 10-item AUDIT. Conclusion: These results strongly support the usefulness of AUDIT-5 for screening of PD, AUD and AD in clinical settings in Korean male populations.