To characterize metabolic effects of troglitazone in type 2 diabetic, obese, and lean subjects, and examine the effects of troglitazone 2-3 weeks after discontinuation. Nine type 2 diabetic, nine obese, and nine lean subjects underwent baseline metabolic studies including an 8-h meal-tolerance test (MTT) and a 5-h glucose clamp. Subjects then received troglitazone (600 mg/day) for 12 weeks and subsequently had repeat metabolic studies. Diabetic subjects remained off hypoglycemic agents for 2-3 weeks and then underwent a 5-h glucose clamp. In diabetic subjects, fasting plasma glucose was reduced (P<0.05) and insulin-stimulated glucose disposal (Rd) was enhanced by treatment (P<0.02). The area under the MTT 8-h plasma glucose curve declined with therapy (P<0.001), and its change was positively correlated with the improvement in Rd (r = 0.75, P<0.05). There was also a positive correlation between the change in fasting hepatic glucose output (HGO) and the change in fasting plasma glucose with treatment (r = 0.92, P<0.001). Discontinuation of therapy for 2-3 weeks did not significantly affect fasting plasma glucose or insulin-stimulated glucose Rd. In obese subjects, insulin-stimulated glucose Rd improved with therapy (P<0.001), allowing for maintenance of euglycemia by lower plasma insulin concentrations (P<0.05). In lean subjects, an increase in fasting HGO (P<0.001) and glucose clearance (P<0.01) was observed. Troglitazone lowers fasting and postprandial plasma glucose in type 2 diabetes by affecting both fasting HGO and peripheral insulin sensitivity. Its effects are evident 2-3 weeks after discontinuation. In obese subjects, its insulin sensitizing effects suggest a role for its use in the primary prevention of type 2 diabetes.