The current study aimed to compare the effects of treatment (2 months) with thiazolidinediones (TZDs) and placebo on glucose and fat metabolism in patients with type 2 diabetes (T2DM) in a crossover design. Eight patients received placebo (2 months) followed by TZD (2 months). Two-stage (1.5 and 6.0 pmol/kg min) hyperinsulinemic-euglycemic clamps were performed in all 8 patients pre- and post-placebo and post-TZD (post-placebo = pre-TZD). We determined rates of glucose disappearance (G_Rd), glycolysis (GLS), glycogen synthesis (GS) (all with 3-³H glucose), carbohydrate (CHO) oxidation (indirect calorimetry), endogenous glucose production (EGP), free fatty acid (FFA) release (²H₅ glycerol), and oxidation (indirect calorimetry) and re-esterification, as well as plasma adiponectin and leptin concentrations, and fat cell size and number (determined in upper thigh biopsy samples). Placebo treatment had no effects on any of the measured parameters. TZD treatment improved insulin-stimulated glucose uptake (ISGU) from 17.1 to 26.4 μmol/kg min (P < .01) and insulin-stimulated GS from 4.8 to 13.4 μmol/kg min (P < 0.03), potentiated insulin-induced suppression of lipolysis from 4.3 to 2.3 μmol/kg min (P < .03) and FFA re-esterification from 1.9 to 1.0 μmol/kg min (P < .02), increased plasma adiponectin levels from 2.7 to 7.2 μg/mL (P < .05), and decreased plasma leptin levels from 30.8 to 23.4 ng/mL (P < .02). In addition, TZD tended to increase the number of small adipocytes (<50 μm) in subcutaneous adipose tissue. We conclude that TZDs have multiple actions and that many, but perhaps not all, can be accounted for by their action on adipose tissue.