Under-reporting of alcohol use by HIV-infected patients could adversely impact clinical care. This study examined factors associated with under-reporting of alcohol consumption by patients who denied alcohol use in clinical and research settings using an alcohol biomarker. We enrolled ART-naïve, HIV-infected adults at Mbarara Hospital HIV clinic in Uganda. We conducted baseline interviews on alcohol use, demographics, Spirituality and Religiosity Index (SRI), health and functional status; and tested for breath alcohol content and collected blood for phosphatidylethanol (PEth), a sensitive and specific biomarker of alcohol use. We determined PEth status among participants who denied alcohol consumption to clinic counselors (Group 1, n = 104), and those who denied alcohol use on their research interview (Group 2, n = 198). A positive PEth was defined as ≥8 ng/ml. Multiple logistic regression models were used to examine whether testing PEth-positive varied by demographics, literacy, spirituality, socially desirable reporting and physical health status. Results showed that, among the 104 participants in Group 1, 28.8% were PEth-positive. The odds of being PEth-positive were higher for those reporting prior unhealthy drinking (adjusted odds ratio (AOR): 4.7, 95% confidence interval (CI): 1.8, 12.5). No other factors were statistically significant. Among the 198 participants in Group 2, 13.1% were PEth-positive. The odds of being PEth-positive were higher for those reporting past unhealthy drinking (AOR: 4.6, 95% CI: 1.8, 12.2), the Catholics (AOR: 3.8, 95% CI: 1.3, 11.0) compared to Protestants and lower for the literate participants (AOR: 0.3, 95% CI: 0.1, 0.8). We concluded that under-reporting of alcohol use to HIV clinic staff was substantial, but it was lower in a research setting that conducted testing for breath alcohol and PEth. A report of past unhealthy drinking may highlight current alcohol use among deniers. Strategies to improve alcohol self-report are needed within HIV care settings in Uganda.