SETTING: Arusha, Mwanza, Mufindi and Kilosa in Tanzania. OBJECTIVE: To assess the test characteristics of three indirect adherence measures against a gold standard of direct measurements of drug intake for use in a resource-constrained setting. METHODS: We assessed sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV), and the diagnostic odds ratio (dORs) of three indirect adherence measurement tools against direct measurement in urine using the IsoScreen assay. RESULTS: The single adherence question of missed doses in the last 2 days had the highest dOR (40.3) compared to the Morisky medication adherence scale (MMAS, 2.5) and pill counts (3.4). The sensitivities of these measures were respectively 97.9%, 92% and 89.6%. Specificity ranged from 46.4% (adherence question) to 17.9% (MMAS). The PPVs of adherence question, pill counts and MMAS were respectively 97.6%, 96.5% and 94.2%, while the NPVs ranged from 50% (adherence question) to 3.1% (MMAS). CONCLUSION: Among several instruments for indirect adherence measure in the routine setting of the Tanzanian National Tuberculosis and Leprosy Programme, a single adherence question was found to have the best discriminatory power. However, the single adherence question might not adequately identify patients who are non-adherent. Confirmatory studies are needed, especially in settings with low adherence rates.