Peripheral insulin levels are determined by beta-cell secretion, insulin sensitivity, and hepatic insulin extraction (HIE). We have previously shown that whereas sulfonylureas reduce insulin extraction, metformin enhances HIE. However, the effects of thiazolidinediones (TZDs) on HIE remain uncertain. Thus, we investigated the potential contribution of hepatic insulin clearance to peripheral insulin levels during rosiglitazone therapy in African Americans with impaired glucose tolerance (IGT) and type 2 diabetes mellitus (DM). The study was composed of 12 first-degree relatives with IGT and 17 patients with newly diagnosed type 2 DM. Nineteen healthy relatives with normal glucose tolerance served as controls. Serum glucose, insulin, and C-peptide, and HIE (C-peptide-insulin molar ratios) were measured at t = 0 and 120 minutes during oral glucose tolerance test (OGTT) in all the subjects. The OGTT was performed before and after 3 months of rosiglitazone therapy (4 mg/d x 4 weeks and >8 mg/d x 8 weeks) in patients with IGT and type 2 DM. Insulin resistance index and beta-cell function were calculated in each subject using homeostasis model assessment (HOMA). Rosiglitazone therapy improved but did not normalize the overall glycemic control in the IGT and type 2 DM groups. After rosiglitazone therapy, the mean serum insulin and C-peptide levels at fasting remained unchanged. However, the 2-hour serum glucose and insulin were lower, whereas serum C-peptide was unchanged during 3 months of rosiglitazone treatment. Mean insulin resistance index of HOMA was reduced by 30% (4.12 +/- 1.95 vs 6.33 +/- 3.54, P < .05) in the type 2 DM group and by 21% (3.78 +/- 2.45 vs 4.81 +/- 3.49, P = NS) in the IGT group. Mean HIE values were significantly lower (70%) in the type 2 DM and IGT groups when compared with the normal glucose tolerance group. At 3 months, basal HIE was not significantly changed by rosiglitazone therapy in IGT and type 2 DM groups when compared with the baseline (0 month). However, rosiglitazone therapy was associated with increased HIE at 2 hours during OGTT by 40% and 30% in the IGT and type 2 DM groups, respectively, from the baseline (0 month) values. Furthermore, HIE inversely correlated with the insulin resistance index of HOMA (r = -.46, P < .05). We conclude that rosiglitazone therapy improved overall glucose tolerance and enhanced insulin sensitivity in patients with IGT and type 2 DM. Although basal HIE remained unchanged, rosiglitazone therapy increased postglucose challenge HIE in African Americans with IGT and type 2 DM. We speculate that TZDs increase insulin clearance or HIE after oral glucose challenge. This study suggests that in addition to insulin sensitization, rosiglitazone may be involved in insulin metabolism. The significance of the increased insulin clearance by TZD therapy remains uncertain and deserves further investigation in patients with insulin resistance and glucose intolerance.