OBJECTIVES: To compare the effectiveness of brand rosiglitazone maleate (BRM) versus generic rosiglitazone HCl (GRH) in patients with type 2 diabetes mellitus, using computerized records of a healthcare organization. METHODS: Comparison of HbA1C reduction in patients starting treatment with either BRM (n=740) or GRH (n=306) in the years 2004-2005. RESULTS: BRM users were older (63.5±11 vs. 61.7±10 years p<0.001) and presented more cardiovascular disorders (38% vs. 25%, p<0.001) with no differences in gender distribution, rates of hypertension or use of concomitant oral hypoglycemic drugs. Use of concomitant insulin was more frequent (17.7% vs. 6.2%, p<0.0001), rates of dispensed rosiglitazone doses > 4mg/d (65.3% vs. 48.5%, p<0.001) and treatment duration was longer (9.3±6.2 vs. 5.2±2.2 months, p<0.001) with the generic formulation. Baseline HbA1C levels were higher (9.0±1.5 vs. 8.6±1.2%, p<0.001) and the absolute decrease in HbA1C levels was greater in the GRH group (-1.2%±1.6% vs. -0.5%±1.7%, p<0.001). On multiple regression analysis, the decrease in HbA1C (dependent variable) was associated mainly with initial HbA1C level (partial r2=0.30). Rosiglitazone formulation (partial r2=0.02), age, treatment duration and concomitant insulin (partial r2=0.006) were all significant but minor predictors, with no effect of rosiglitazone daily dose. Mean regression-predicted decreases in HbA1C (with 95% CL) were not significantly different between the two rosiglitazone formulations: -1.6% (-4.3% to +1.1%) for GRH and -1.1% (-3.8% to +1.6%) for BRM. CONCLUSION: In this retrospective database analysis, we found no evidence of different effectiveness of generic vs. brand rosiglitazone in lowering HbA1C levels.