Purpose: To compare patient outcomes (diabetes medication compliance, health care costs and utilization) among cohorts choosing conventional (CP) and consumer-directed health plans (CDHP) in a large national employer's diabetes population. Methods: We retrospectively examined medical and pharmaceutical claims data from April 2002 through March 2006 for employees and their dependents. Members ≥18 years of age were included in the analyses if they had at least one diagnosis (ICD-9 code 250.xx) or prescription for diabetes, and ≥six months of claims data following first plan enrollment date. Propensity scoring was used to match members selecting a CP to the CDHP cohort in a one to one ratio. Members were matched on age, gender, length of employment, and geographic region. Adherence to diabetes medications were calculated using the Medication Possession Ratio (MPR, total days of prescription fills divided by 180 days). Categorical medication adherence is defined as an MPR of ≥80%. Analysis of covariance assessed the association between patient outcomes and plan selection in the six month period following the initial fill of a diabetes medication. Results: A total of 1306 members (653 for each cohort) were included in the analysis. Mean age was 42.8 (SD=12.4); 52% were female, with no statistically significant demographic and clinical characteristics differences among the cohorts. Mean MPR for any diabetes drug was significantly lower for the CDHP cohort (0.66) as compared to the CP cohort (0.70, p=0.02). The CDHP cohort had a significantly higher number of outpatient visits (5.2±0.3 vs. 4.0±0.3, p =0.001) and lower prescription costs (697.4±57.5 vs. 1112.6±53.4, p<0.0001) than the CP cohort. Total medical cost was not significantly different between cohorts (4765.5±755.5 vs. 3542.9±700.67, p=0.24). Conclusion: A four percent difference in medication adherence in the first six months of therapy among CDHP and CP cohorts should be considered important, as there is evidence in the literature that the first six months of medication taking is predictive of future behavior. The short monitoring period of this study (180 days post CDHP selection) constrains our ability to draw conclusions regarding utilization, although it is notable that the CDHP cohort had higher utilization of outpatient visits. A longer monitoring period for diabetes medication use following CDHP enrollment is necessary to determine the relationship between plan selection and health outcomes.