OBJECTIVE: Although exercise is recommended as a cornerstone of treatment for type 2 diabetes, it is often poorly adopted by patients. We have noted that even in the absence of apparent cardiovascular disease, persons with type 2 diabetes have an impaired ability to carry out maximal exercise, and the impairment is correlated with insulin resistance and endothelial dysfunction. We hypothesized that administration of a thiazolidinedione (TZD) agent would improve exercise capacity in type 2 diabetes. RESEARCH DESIGN AND METHODS: Twenty participants with uncomplicated type 2 diabetes were randomly assigned in a double-blind study to receive either 4 mg/day of rosiglitazone or matching placebo after baseline measurements to assess endothelial function (brachial artery diameter by brachial ultrasound), maximal oxygen consumption (VO(2max)), oxygen uptake (VO(2)) kinetics, and insulin sensitivity by hyperinsulinemic-euglycemic clamp. Measurements were reassessed after 4 months of treatment. RESULTS: Participant groups did not differ at baseline in any measure. Rosiglitazone-treated participants (n = 10) had significantly improved VO(2max) (19.8 +/- 5.3 ml . kg(-1) . min(-1) before rosiglitazone vs. 21.2 +/- 5.1 ml . kg(-1) . min(-1) after rosiglitazone, P < 0.01), insulin sensitivity, and endothelial function. A change in VO(2max) correlated with improved insulin sensitivity measured by clamp (r = 0.68, P < 0.05) and with improved brachial artery diameter (r = 0.70, P < 0.05). Placebo-treated participants (n = 10) showed no changes in VO(2max) (19.4 +/- 5.2 ml . kg(-1) . min(-1) before rosiglitazone vs. 18.1 +/- 5.3 ml . kg(-1) . min(-1) after rosiglitazone, NS) or brachial artery diameter. CONCLUSIONS: This is the first known report showing that a TZD improved exercise function in type 2 diabetes. Whether this is due to the observed improvements in insulin sensitivity and/or endothelial function or to another action of the TZD class requires further exploration.