Background: The Alcohol Use Disorders Identification Test (AUDIT) is frequently recommended as case-finding instrument for unhealthy alcohol use. However, it consists of ten questions requiring considerable time resources. Using a three-item version (AUDIT-C) yielded promising results in some studies. Objectives: The primary aim of the present study was to systematically review the evidence comparing the screening properties of the AUDIT-C and the full AUDIT for unhealthy alcohol use. The secondary objective was to test the applicability of the difference in area under the receiver operating characteristics curve (dAUC) as effect measure in a comparative diagnostic accuracy meta-analysis. Methods: In December 2007, the MEDLINE, EMBASE, CINAHL, Web of Science, PsycINFO and BIOSIS Previews databases were searched. Three reviewers decided independently about study inclusion. Five possible reference standards were examined: 1) risky drinking; 2) alcohol abuse or harmful drinking; 3) alcohol dependence; 4) alcohol use disorder, defined as the presence of alcohol abuse/harmful drinking or alcohol dependence; and 5) unhealthy alcohol use, defined as the presence of risky drinking and/or any alcohol use disorder. Methodological quality was assessed by a checklist derived form the QUADAS tool. AUC-differences were pooled using inverse variance weighted random-effects model. Statistical heterogeneity was quantified using the I² statistic. Results: A total of 15 studies were included in the meta-analysis. In case of screening for risky drinking, alcohol abuse or harmful drinking, and unhealthy alcohol use, the full AUDIT did not prove to be significantly superior to the AUDIT-C (dAUC = 0.002, -0.001, and -0.002, respectively). In case of screening for alcohol dependence and alcohol use disorder, the AUDIT showed a statistically significant advantage over the AUDIT-C (dAUC = -0.035 and -0.025, respectively). Conclusions: The AUDIT-C is not inferior to the full AUDIT as a screening instrument for risky drinking alone or in combination with any alcohol use disorder and thus may be used as an economic casefinding instrument. The method of using difference in AUCs as effect measure proved to be applicable for comparing diagnostic instruments with multiple thresholds.