OBJECTIVE: The goal of this study was to determine the validity and utility of a pharmacy-based time-to-refill measure of antiretroviral therapy adherence. METHODS: We performed an observational cohort study of 110 HIV-infected subjects on a stable, highly active antiretroviral regimen for at least 3 months at a Veterans' Affairs Medical Center in Philadelphia, Pennsylvania. RESULTS: The viral load decreased by 0.12 log c/mL (95% confidence interval [CI] 0.01-0.23 log c/mL) for each 10% increase in pharmacy-based time-to-refill defined adherence as compared with 0.05 log c/mL (95% CI -0.14-0.25 log c/mL) for the self-reported adherence measure. Thus, only the refill-defined measure was statistically significantly associated with viral load change. When adherence was classified as good (> or = 85%) versus poor (<85%), both measures demonstrated a significant difference in outcome between groups. Yet, in individuals self-reporting 100% adherence, those classified as good adherers using the pharmacy-based measure had greater viral load reductions than poor adherers (2.4 log c/mL [interquartile range 1.4-3.4] vs. 1.5 log c/mL [interquartile range 0.8-2.4, P=.03). CONCLUSIONS: The pharmacy-based technique is a valid measure of antiretroviral therapy adherence. Because it provides clinically relevant information in subjects who self-report 100% adherence, it should be incorporated into clinical practice and adherence research.