We studied the effects of troglitazone (200-800 mg daily) or placebo on carbohydrate metabolism in 18 subjects with type 2 diabetes (mean age, 66 yr; body mass index, 27.7 kg/m2) at baseline and after taking medication for 12 weeks. We measured fasting proinsulin (PI) and immunoreactive insulin (IRI) levels in all subjects. Thirteen subjects underwent additional metabolic studies, including injection of arginine to determine the acute insulin response, and an i.v. glucose tolerance test to measure the insulin sensitivity index (SI) and glucose effectiveness at zero insulin using the minimal model, i.v. glucose tolerance, and acute insulin response to glucose. Troglitazone treatment resulted in a decrease in fasting plasma glucose from 11.2 +/- 0.7 to 9.6 +/- 0.9 mmol/L (P = 0.02). This was associated with a decrease in the fasting IRI concentration (111 +/- 20 to 82 +/- 13 pmol/L; P = 0.02) and a trend toward a decrease in the fasting PI concentration (43 +/- 11 to 25 +/- 4 pmol/L; P = 0.06). A significant decrease in PI/IRI was observed (38.3 +/- 3.6% to 32.6 +/- 3.2%; P = 0.04). Troglitazone therapy was also associated with a decrease in the acute insulin response to arginine (226 +/- 34 to 167 +/- 25 pmol/L; P = .01) and a near-significant percent increase in S(I) (75 +/- 35%; P = 0.06). Glucose effectiveness at zero insulin, i.v. glucose tolerance, and acute insulin response to glucose did not change. Thus, we found that the decrease in plasma glucose during troglitazone therapy is associated with a dose-related decrease in PI/IRI and an increase in S(I), suggesting that changes in both B cell function and insulin sensitivity contribute to the improvement in metabolic status.