OBJECTIVE: The purpose of this study was to conduct a psychometric evaluation of the Patient Medication Adherence Questionnaire (PMAQ), an HIV-specific instrument designed to assess medication-taking behavior as well as barriers and motivators affecting adherence. METHODS: The PMAQ was prospectively administered to 410 patients in 3 studies: a) a focus group of HIV infected patients (n=67); b) a cross-sectional observational cohort study (n=120); and c) a clinical trail, with evaluations at baseline, and weeks 8 and 16 (n=223). The Multitrait/Multi-item Analysis Program-Revised was used to evaluate reliability and validity for each of the 5 hypothesized PMAQ dimensions: Remembering (REM), Scheduling and Timing (SCH), Physical Effects(PE), Knowledge and Attitudes(KNO), and Social Support(SS). RESULTS: Results consistent across the studies and across multiple visits within the clinical trial are reported. Cronbach’s alpha for 4 dimensions ranged between 0.43 to 0.73 and between 0.76 to 0.79 for the total score. For the same dimensions, discriminant validity was sufficient to suggest multitrait scaling even though a few items within the dimensions produced weak (<0.40) item-scale correlations. To assess evidence of convergent validity, the same dimensions had scale-scale correlations > 0.42. The dimension that did not perform well across these studies was KNO. Given its importance in earlier research, this finding most probably reflects the heterogeneous nature of the items within this dimension. The correlations between REM, SCH, and PE with missed doses were moderate. SS was unrelated to missed doses. CONCLUSION: Even though the KNO dimension warrants further evaluation, the PMAQ generally demonstrates adequate psychometric properties. It can be used to assess barriers and motivators to adherence in research settings, and may be predictive of adherence.