OBJECTIVE Surveys assessing alcohol use among physicians have most commonly employed the Alcohol Use Disorders Identification Test (AUDIT) or the AUDIT-C, the most common short version of the AUDIT. As with other screeners, prevalence estimation is dependent on the accuracy of the test as well as choice of the cutoff value. The aim of the current study is to derive more precise prevalence estimates of alcohol problems in physicians by correcting for false-positive and false-negative results. METHOD In the context of a survey, the AUDIT was sent out via email or standard postal service to all 2484 physicians in Salzburg, Austria. A total of 456 physicians participated. A published correction formula was used to estimate the real prevalence of alcohol use problems. RESULTS Applying a cutoff of 5 points for the AUDIT-C, 15.7% of female and 37.7% of male physicians screened positive. Use of a correction based on general population data and the sensitivity and specificity of the AUDIT-C resulted in much lower prevalence rates: 4.0% for female and 9.5% for male physicians. Using the full AUDIT, 19.6% of the female physicians and 48% of the male physicians were screened positive. Using the correction, the estimated prevalence rates for females and males were 6.3% and 15.5%, respectively. CONCLUSIONS Our findings demonstrate that uncorrected screening results may markedly overestimate the prevalence of physicians drinking problems.