The insulin-sensitizing effects of thiazolidinediones are thought to be mediated through peroxisome proliferator-activated receptor-γ, a nuclear receptor that is highly abundant in adipose tissue. It has been reported that adipocytes secrete a variety of proteins, including tumor necrosis factor-α, resistin, plasminogen activator inhibitor-1, and adiponectin. Adiponectin is a fat cell-secreted protein that has been reported to increase fat oxidation and improve insulin sensitivity. Our aim was to study the effects of troglitazone on adiponectin levels in lean, obese, and diabetic subjects. Ten diabetic and 17 nondiabetic subjects (8 lean, BMI <27 kg/m² and 9 obese, BMI >27 kg/m²) participated in the study. All subjects underwent an 80 mU · m⁻² · min⁻¹ hyperinsulinemic-euglycemic glucose clamp before and after 3 months’ treatment with the thiazolidinedione (TZD) troglitazone (600 mg/day). Fasting plasma glucose significantly decreased in the diabetic group after 12 weeks of treatment compared with baseline (9.1 ± 0.9 vs. 11.1 ± 0.9 mmol/l, P < 0.005) but was unchanged in the lean and obese subjects. Fasting insulin for the entire group was significantly lower than baseline (P = 0.02) after treatment. At baseline, glucose disposal rate (R_d) was lower in the diabetic subjects (3.4 ± 0.5 mg · kg⁻¹ · min⁻¹) than in the lean (12.3 ± 0.4) or obese subjects (6.7 ± 0.7) (P < 0.001 for both) and was significantly improved in the diabetic and obese groups (P < 0.05) after treatment, and it remained unchanged in the lean subjects. Baseline adiponectin levels were significantly lower in the diabetic than the lean subjects (9.0 ± 1.7 vs. 16.7 ± 2.7 μg/ml, P = 0.03) and rose uniformly in all subjects (12.2 ± 2.3 vs. 25.7 ± 2.6 μg/ml, P < 10^–4) after treatment, with no significant difference detected among the three groups. During the glucose clamps, adiponectin levels were suppressed below basal levels in all groups (10.2 ± 2.3 vs. 12.2 ± 2.3 μg/ml, P < 0.01). Adiponectin levels correlated with R_d (r = 0.46, P = 0.016) and HDL cholesterol levels (r = 0.59, P < 0.001) and negatively correlated with fasting insulin (r = −0.39, P = 0.042) and plasma triglyceride (r = −0.61, P < 0.001). Our findings show that TZD treatment increased adiponectin levels in all subjects, including normal subjects in which no other effects of TZDs are observed. Insulin also appears to suppress adiponectin levels. We have confirmed these results in normal rats. These findings suggest that adiponectin can be regulated by obesity, diabetes, TZDs, and insulin, and it may play a physiologic role in enhancing insulin sensitivity.