OBJECTIVES: The objective is to compare patterns of patient compliance on a fixed-dose combination therapy (FDCT) versus single agent combination therapy (SACT) with any newly initiated thiazolidinedione (TZD) and metformin (MET) for the treatment of diabetes. METHODS: This study utilized a retrospective pharmacy claims database of continuously eligible patients who were newly initiated on a TZD or MET either in FDCT or SACT. Patients during December 2005 to June 2006 were identified and followed for six months. Patients who switched between FDCT and SACTs were excluded. Due to a national Avandamet shortage spanning July 2005 to June 2006, patients on Avandamet were excluded resulting in comparison between Actoplus Met as FDCT and SACT groups. Adherence was defined as proportion of days covered when both TZD and MET were available >80% of the time. Propensity-score weighted logistic regression was used to adjust for patient demographics and plan characteristics. RESULTS: There were 729 and 19,440 patients on FDCT and SACT respectively. Mean age were 55 and 58 years (P<0.0001), proportion female were 39% and 43% (P<0.0362), and prior sulfonylurea utilization rates at baseline were 30% and 53% (P<0.0001), respectively. Distribution of the patient population in various market segments (managed care, Medicaid, Medicare, and Self-Insured) differed. Unadjusted adherence rates were 51% and 48% for FDCT and SACT respectively. Adjusting for baseline characteristics, patients on FDCT were 50% more likely to be adherent to therapy [OR=1.50, 95% CI= 1.46 – 1.55]. CONCLUSION: When controlling for multiple factors, patients on FDCT were more likely to be adherent to therapy compared to those on SACT.