Fat redistribution in the setting of protease inhibitor use is increasingly common and is associated with insulin resistance in human immunodeficiency virus (HIV)-infected patients. However, little is known regarding the factors that may contribute to abnormal insulin regulation in this population. We assessed fasting insulin levels in HIV-infected men and determined the relationship among insulin, body composition, endogenous gonadal steroid concentrations, and antiviral therapy in this population. We also determined the effects of exogenous testosterone administration using the homeostatic model for insulin resistance (HOMA IR) in hypogonadal HIV-infected men with the acquired immunodeficiency syndrome wasting syndrome. Fifty HIV-infected men with acquired immunodeficiency syndrome wasting were compared with 20 age- and body mass index (BMI)-matched healthy control subjects. Insulin concentrations were significantly increased in HIV-infected patients compared to those in control patients (16.6+/-1.8 vs. 10.4+/-0.8 microU/mL; P<0.05) and were increased in nucleoside reverse transcriptase (NRTI)-treated patients who did not receive a protease inhibitor (PI; 21.7+/-4.3 vs. 10.4+/-0.8 microU/mL; P<0.05). Insulin concentrations and HOMA IR were inversely correlated with the serum free testosterone concentration (r = -0.36; P = 0.01 for insulin level; r = -0.30; P = 0.03 for HOMA), but not to body composition parameters, age, or BMI. In a multivariate regression analysis, free testosterone (P = 0.05), BMI (P<0.01), and lean body mass (P = 0.04) were significant. Lower lean body mass and higher BMI predicted increased insulin resistance. The HIV-infected patients demonstrated an increased trunk fat to total fat ratio (0.49+/-0.02 vs. 0.45+/-0.02; P<0.05) and an increased trunk fat to extremity fat ratio (1.27+/-0.09 vs. 0.95+/-0.06, P = 0.01), but a reduced extremity fat to total fat ratio (0.44+/-0.01 vs. 0.49 + 0.01; P = 0.02) and reduced overall total body fat (13.8+/-0.7 vs. 17.2+/-0.9 kg; P<0.01) compared to the control subjects. Increased truncal fat and reduced extremity fat were seen among NRTI-treated patients, but this pattern was most severe among patients receiving combined NRTI and PI therapy [trunk fat to extremity ratio, 1.47+/-0.15 vs. 0.95+/-0.06 (P<0.01); extremity fat to total fat ratio, 0.40+/-0.02 vs. 0.49+/-0.01 (P<0.05)]. Insulin responses to testosterone administration were investigated among 52 HIV-infected men with hypogonadism and wasting (weight <90% ideal body weight and/or weight loss >10%) randomized to either testosterone (300 mg, im, every 3 weeks) or placebo for 6 months. Testosterone administration reduced HOMA IR in the HIV-infected men (-0.6+/-0.7 vs. +1.41+/-0.8, testosterone vs. placebo, P = 0.05) in association with increased lean body mass (P = 0.02). These data demonstrate significant hyperinsulinemia in HIV-infected patients, which can occur in the absence of PI use. In NRTI-treated patients not receiving PI, a precursor phenotype is apparent, with increased truncal fat, reduced extremity fat, and increased insulin concentrations. This phenotype is exaggerated in patients receiving PI therapy, with further increased truncal fat and reduced extremity fat, although hyperinsulinemia per se is not worse. Endogenous gonadal steroid levels are inversely related to hyperinsulinemia in HIV-infected men, but reduced lean body mass and increased weight are the primary independent predictors of hyperinsulinemia. Indexes of insulin sensitivity improve in response to physiological androgen administration among hypogonadal HIV-infected patients, and this change is again related primarily to increased lean body mass in response to testosterone administration.