<sec><title>BACKGROUND</title>Chemotherapy to prevent TB is a core component of care for persons living with HIV (PLHIV). There are few reports describing adherence to TB prevention under programmatic conditions in high TB burden settings.</sec><sec><title>METHODS</title>We measured adherence to daily isoniazid (INH) preventive treatment (IPT) using a commercially available colourimetric assay to detect urine INH metabolites among PLHIV who self-reported INH ingestion within the preceding 24 h. Enrollee characteristics associated with non-adherence despite self-reported INH ingestion were identified in multivariate log-binomial regression. Interoperator reliability for the detection of INH metabolites was calculated among three independent operators.</sec><sec><title>RESULTS</title>Self-reported INH ingestion and metabolite data were known for 300 PLHIV. INH metabolite was detected in 112 (68.7%) of 163 PLHIV who self-reported INH ingestion in the preceding 24 h. The prevalence of alcohol consumption was significantly higher among INH-non-adherent PLHIV compared with INH-adherent PLHIV (adjusted prevalence ratio 2.43, 95% CI 1.16-5.12). Two-way interoperator reliability ranged from κ 0.86 to κ 0.94.</sec><sec><title>CONCLUSIONS</title>Compared with self-reported 24-h INH ingestion in a high TB-HIV-incidence programmatic setting, biometric adherence to IPT was suboptimal in this sample of PLHIV. Alcohol consumption was the only potentially modifiable risk factor significantly associated with INH non-adherence. Colourimetric interpretation reliability across three operators was moderate/strong.</sec>.