Background: There have been a number of studies relating medication adherence to patient characteristics. There is less research on influence of health care providers on patients’ medication-taking behavior. Objectives: To evaluate the pharmacy-level effect on medication adherence for patients receiving antidiabetic medications. Research Design: This was a hypothesis-driven retrospective study using cross-sectional design and insurance claims data. The main analytical interest was the pharmacy-level effect on proportion of days covered as the measure of medication adherence. Multilevel random and mixed-effect models were used to tease out the pharmacy-level effect on patient outcomes. Subjects: The study population consisted individuals aged 18–64 years, insured under employer-sponsored private health plans. Results: We estimated models with and without covariates. In both models, pharmacy cluster effect was statistically significant (Pqqqlt;0.001). In the model without covariates, pharmacy cluster effect accounted for 12.8% (95% confidence interval, 12.4%–13.1%) of total variance in adherence, whereas in the model with covariates pharmacies accounted for 12.1% (95% confidence interval, 11.6%–12.4%) of total variance. Covariates associated significantly with adherence were age, sex, mail order pharmacy, and prescription drug copay. Conclusions: The results suggest significant variation in medication adherence attributable to pharmacy factor, independent of other effects. The underlying reason could be varying level of influence from pharmacies’ efforts to inform or influence patients to take medications in prescribed manners. More research is necessary to better understand the effect of specific pharmacy characteristics and practice styles differences.