BACKGROUND: Good adherence is essential for successful antiretroviral therapy (ART) provision, but simple measures have rarely been validated in Africa. METHODS: This was an observational analysis of an open multicenter randomized HIV/AIDS management trial in Uganda and Zimbabwe. At 4-weekly clinic visits, ART drugs were provided and adherence measured through pill usage and questionnaire. Viral load response was assessed in a subset of patients. Drug possession ratio (percentage of drugs taken between visits) defined complete (100%) and good (>or=95%) adherence. RESULTS: In 2,957 patients, 90% had pill counts at every visit. Good adherence increased from 87%, 4 weeks after ART initiation, to 94% at 48 weeks, but only 1,454 (49%) patients achieved good adherence at every visit in the first year. Complete adherence was associated with 0.32 greater reduction in log10 viral load (95% confidence interval 0.05, 0.60 P = 0.02) and was independently associated with higher baseline CD4 count, starting ART later in the trial, reporting a single regular sexual partner, clinical center, and time on ART. CONCLUSIONS: Population level adherence improved over time suggesting an association with clinical experience. Most patients had at least one visit in the year on which they reported not having good adherence, showing the need for continued adherence interventions.