OBJECTIVE: Impaired endothelial function was demonstrated in HIV-infected persons on protease inhibitor (PI)-containing antiretroviral therapy, probably due to altered lipid metabolism. Atazanavir is a PI causing less atherogenic lipoprotein changes. This study determined whether endothelial function improves after switching from other PI to atazanavir. DESIGN: Randomised, observer-blind, treatment-controlled trial. SETTING: Three university-based outpatient clinics. PATIENTS: 39 HIV-infected persons with suppressed viral replication on PI-containing regimens and fasting low-density lipoprotein (LDL)-cholesterol greater than 3 mmol/l. INTERVENTION: Patients were randomly assigned to continue the current PI or change to unboosted atazanavir. MAIN OUTCOME MEASURES: Endpoints at week 24 were endothelial function assessed by flow-mediated dilation (FMD) of the brachial artery, lipid profiles and serum inflammation and oxidative stress parameters. RESULTS: Baseline characteristics and mean FMD values of the two treatment groups were comparable (3.9% (SD 1.8) on atazanavir versus 4.0% (SD 1.5) in controls). After 24 weeks' treatment, FMD decreased to 3.3% (SD 1.4) and 3.4% (SD 1.7), respectively (all p = ns). Total cholesterol improved in both groups (p<0.0001 and p = 0.01, respectively) but changes were more pronounced on atazanavir (p = 0.05, changes between groups). High-density lipoprotein and triglyceride levels improved on atazanavir (p = 0.03 and p = 0.003, respectively) but not in controls. Serum inflammatory and oxidative stress parameters did not change; oxidised LDL improved significantly in the atazanavir group. CONCLUSIONS: The switch from another PI to atazanavir in treatment-experienced patients did not result in improvement of endothelial function despite significantly improved serum lipids. Atherogenic lipid profiles and direct effects of antiretroviral drugs on the endothelium may affect vascular function. Trial registration number: NCT00447070.