INTRODUCTION: Valid Alcohol Use Disorder (AUD) and Substance Use Disorder (SUD) screeners are needed to identify and link people to services. We evaluated the performance of several AUD and SUD screeners in South Africa using the Mini International Neuropsychiatric Interview (MINI)-5 diagnostic gold standard. METHODS: Adults at primary and tertiary care facilities in Buffalo City Metro, South Africa, were screened by research assistants using the AUDIT and AUDIT-C (AUD), DAST-10 (SUD) and NIDA Quick Screen (AUD and SUD). Nurses administered the MINI-5 to identify AUD and SUD. We assessed the internal consistency, criterion validity, sensitivity and specificity of these tools, stratified by gender. RESULTS: Among 1885 participants, the prevalence of AUD and SUD were 9.5 % and 1.6 %, respectively. All tools demonstrated adequate internal consistency and criterion validity. A positive AUDIT screen (men: ≥8
women: ≥7) yielded sensitivity/specificity of 70.6/87.3 % (men: 78.7/82.6 %
women: 64.8/89.8 %). A positive AUDIT-C screen (men: ≥4
women: ≥3) yielded sensitivity/specificity of 66.1/82.0 % (men: 64.0/78.8 %
women: 67.6/81.3 %). Endorsing the NIDA alcohol use question yielded sensitivity/specificity of 71.1/68.1 % (men: 74.7/59.7 %
women: 68.6/72.5 %). Endorsing either NIDA substance use questions yielded sensitivity/specificity of 80.6/91.7 % (men: 80.8/89.0 %
women: 80.0/93.1 %). A DAST-10 cut-off of ≥ 3 yielded sensitivity/specificity of 71.0/96.0 % (men: 73.1/83.7 %
women 60.0/97.4 %). CONCLUSIONS: The AUDIT and AUDIT-C performed similarly among men and women, although lower cut-offs may optimize performance among women. The low number of SUD cases hampered our ability to draw conclusions about the SUD screeners' performance.