We recently reported that treatment with a pharmacologic dose of recombinant human growth hormone (GH) resulted in a significant loss of body fat and gain in lean tissue in HIV-infected patients with syndromes of fat accumulation. However, insulin-mediated glucose disposal decreased transiently after one month of GH therapy. The present paper focuses on the changes of hepatic carbohydrate and fat metabolism associated with GH treatment in the same subjects. We assessed hepatic insulin sensitivity under both fasting and hyperinsulinemic-euglycemic clamp conditions prior to and after one and six months of GH treatment (3 mg/day) in five patients using stable isotope tracer techniques. Indirect calorimetry, and measurements of lipid concentrations. Fasting endogenous glucose production (EGP) increased significantly at one month (12.0 +/- 0.7 to 14.9 +/- 0.9 micromol/kg/min, P < 0.03), and the increase was sustained at six months of GH treatment (14.0 +/- 1.1 micromol/kg/min, NS). This increase in EGP was driven in part by increased glucogenesis (GNG) (3.5 +/- 0.9 to 5.2 +/- 0.9 and 5.8 +/-1.2 micromol/kg/min, n = 4, P < 0.01 and P < 0.01 at one and six months, respectively); small changes in hepatic glycogenolysis also contributed. Sustained increases in lipolysis and progressive decreases in hepatic fractional de novo lipogenesis (DNL) and triglyceride concentrations occurred with GH treatment. These changes were accompanied by an improved lipid profile with a significant increase in HDL cholesterol and significant decreases in total and LDL cholesterol and triglyceride levels, the latter consistent with the decrease in hepatic DNL. During a hyperinsulinemic-euglycemic glucose clamp, EGP and GNG were markedly suppressed compared to the corresponding time points under fasting conditions, albeit less so when measured after one month of GH treatment. Thus, in HIV-infected patients with abnormal fat distribution, pharmacologic doses of GH improved the overall lipid profile, but worsened glucose homeostasis under both fasting and hyperinsulinemic conditions. The combined implications of these positive and negative metabolic effects for cardiovascular disease risk remain unknown.