Although protease inhibitor (PI) therapy has improved the clinical status of patients with HIV infection, concerns have arisen that such treatment may have deleterious effects on glucose control, lipid metabolism, and body fat distribution. To determine whether initiation of PI therapy uniquely affects glucose and lipid metabolism, we analyzed paired data in HIV-infected patients before and after beginning antiretroviral therapy that included a PI (PI; N= 20) or lamivudine (3TC) but no PI (3TC; N= 9); and a control group on stable regimens that included neither of these agents (CONT; N= 12). Measurements included fasting glucose; insulin; triglycerides; total, high-density lipoprotein (HDL) and low-density lipoprotein (LDL) cholesterol; HIV RNA; CD4+ lymphocytes; weight; and total and regional body composition. Neither weight nor total or regional fat content changed significantly in any group during the period of observation. Nonetheless, in patients beginning PI therapy, there were significant increases in glucose (+ 9+/-3 mg/dl; p=. 0136), insulin (+ 12.2+/-4.9 U/ml; p=. 023), triglycerides (+ 53+/-17 mg/dl; p=. 0069), and total and LDL cholesterol (+ 32+/-11 and+ 18+/-5 mg/dl; p=. 0082 and. 0026, respectively). None of these parameters changed significantly in the 3TC or CONT groups. The PI and 3TC groups experienced comparable increases in CD4+ lymphocytes, suggesting that the observed metabolic effects may be associated with PIs uniquely, rather than improvement in clinical status. However, it is also possible that the metabolic effects of PIs are associated with more effective viral suppression, because a greater proportion of patients in the PI group achieved undetectable levels of virus. We conclude that changes in glucose and lipid metabolism are induced by PI therapy in the absence of significant changes in weight or fat distribution. Longer periods of follow-up will be required to determine the clinical significance of these changes. However, at the moment, the risks associated with these metabolic effects do not appear to outweigh improvements in survival seen with PI therapy.