There is increasing interest in home based testing and treatment of HIV to expand access to treatment in sub-Saharan Africa. Such programs rely on self-reported HIV history and use of antiretroviral therapy (ART). However, the accuracy of self-reported ART use in community settings is not well described. In this study, we compared self-reported ART (SR-ART) use in a home based survey against biological exposure to ART (BE-ART), in a population study of older adults in South Africa. Health and Aging in Africa: a Longitudinal Study of an INDEPTH community in South Africa (HAALSI) is a cohort of adults aged 40 +. The baseline home-based interview included self-reported HIV status and ART use. All participants also underwent biological testing for HIV antibodies, viral load and exposure to emtricitabine (FTC) or lamivudine (3TC), which are included in all first-line and second-line ART regimens in the public-sector South African HIV program. We calculated the performance characteristics for SR-ART compared to BE-ART and fit multivariable logistic regression models to identify correlates of invalid SR-ART responses. Of 4560 HAALSI participants with a valid HIV test result available, 1048 (23%) were HIV-positive and 734 [70% of people living with HIV (PLWH)] were biologically validated ART users (BE-ART). The sensitivity of SR-ART use was 64% (95% CI 61-68%) and the specificity was 94% (95% CI 91-96%); the positive predictive value (PPV) was 96% (95% CI 94-98%) and negative predictive value (NPV) was 52% (95% CI 48-56%). We found no sociodemographic predictors of accurate SR-ART use. Over one in three individuals with detectable ART in their blood denied current ART use during a home-based interview. These results demonstrate ongoing stigma related to HIV and its treatment, and have important implications for community health worker programs, clinical programs, and research studies planning community-based ART initiation in the region.