Background: Antiretroviral adherence is essential to HIV treatment efficacy. Various self-reported measures are commonly used for assessing antiretroviral adherence. Limited data are available regarding the validity of those selfreported measures in comparison with long-term objective biomarkers of adherence measures such as hair measures. Methods: Self-reported adherence (frequency, percentage, and visual analog scale [VAS]) and hair tenofovir concentration were evaluated at a single time point from 268 people living with HIV in China. The responses to each of three self-reported measures were converted into percentage and then dichotomized as "optimal" (100%) vs. "suboptimal" (less than 100%) adherence. Two composite adherence scores (CAS) were created from the three self-reported measures: (1) an overall adherence was the average percentage of the three self-reported measures; (2) responses were termed optimal adherence if participants reporting optimal adherence in all three self-reported measures, while were termed suboptimal adherence. Hair tenofovir concentration was also dichotomized as "optimal" (above the limit of quantitation, 36 pg/mg) vs. "suboptimal" adherence (blow 36 pg/mg). Spearman correlation, kappa statistics, and logistic regression analysis were used to calculate the correlations, agreements, and predictions of self-reported measures with hair measure, respectively. Results: Overall adherence, but any of the three self-reported adherence, was correlated with hair tenofovir concentration (r = 0.13, p < 0.05). Self-reported optimal adherence in VAS and CAS measures were agreed with and predicted optimal adherence assessed by hair measure (Kappa = 0.107, adjusted OR = 1.88, 95% CI 1.03-3.45; Kappa = 0.109, adjusted OR = 1.80, 95% CI 1.02-3.18; all p < 0.05, respectively). Conclusion: VAS may be a good individual self-reported measure for antiretroviral adherence, and CAS may be a good composite self-reported measure for antiretroviral adherence.