OBJECTIVES: To use claims data to estimate the association between adherence to medications and clinical outcomes in DM2 patients. METHODS: Using automated data from a large health maintenance organization, we identified a sample of 644 patients age ?18 years diagnosed with DM2, hypertension, and hypercholesterolemia who receive treatment with either cardiovascular or antidiabetic medications between 1999-2001. Patients taking insulin were excluded. We measured adherence to three classes of drugs, biguanides, HMG-CoA reductase inhibitors (CoA-I), and ACE-inhibitors (ACE). Within each drug class, adherence was measured for those patients who filled at least one prescription per year. Body mass index and blood pressure measurements were obtained by medical record ion. Non-adherence was calculated as a continuous measure of medication gaps (CMG), which is the proportion of days without medication during the period of observation. T-tests or non-parametric tests were used, where appropriate, to compare baseline characteristics and differences in clinical outcomes. Associations between clinical outcomes and CMG were estimated using correlation coefficients. Linear regression models were used to adjust for socio-demographic and clinical characteristics. RESULTS: Correlation coefficients were 0.25 for non-adherence to biguanides and HbA1c (p <0.01), 0.32 for non-adherence to CoA-I and LDL (p <0.01), and 0.16 for non-adherence to ACE and SBP (p <0.01). When comparing adherent patients (1-CMG ??80%) to non-adherent patients (<80%), mean ± SD levels of HbA1c, LDL, and BP were lower in the adherent than in the non-adherent group: 8.0 ± 1.2 vs. 8.5 ± 1.4 % for HbA1c (p <0.01), 103.2 ± 28.1 vs. 123.3 ± 34.2 mg/dL for LDL (p <0.01); and 137.8 ± 12.4 vs. 142.9 ± 12.9 mmHg for SBP (p<0.01). Adjustment for socio-demographic characteristics did not attenuate the strength of the associations. CONCLUSIONS: Adherence measured by claims data correlated with clinical outcomes in DM2 patients.