To evaluate clinical, immunologic, and virologic performance of patients with nadir CD4 counts of >350 cells/microL upon treatment interruption. Randomized, open-label clinical trial of 48 weeks' duration. Patients on effective highly active antiretroviral therapy, with nadir CD4 counts of >350 cells/microL and peak viral loads of <50,000 copies/mL were randomized to continue therapy or to interrupt antiretroviral medication. End points for patients with treatment interruption were CD4 counts of <350 cells/microL, viral loads of >1 log above the pretherapy values, or clinical symptoms attributable to HIV, at which point treatment was restarted. In the continuation group, the end points were virologic failure, opportunistic infections, and treatment discontinuation due to toxicities. Twenty patients were randomized to stop therapy and 16 patients to continue. Median CD4 counts at baseline were 643 cells/microL for the interruption group and 633 cells/microL for the continuation group. No end points were reached in the interruption group. By week 8, viral load returned to values comparable to those of pretherapy in all patients in the interruption group and remained stable until week 48. CD4 counts dropped in the interruption group (median loss of 156 cells/microL) at week 48. Significant decreases in venous lactate were observed in the interruption group. Treatment interruptions in patients with nadir CD4 counts of >350 cells/microL seem safe for at least 48 weeks. Pretherapy viral load appears as a valuable tool to predict its level at week 48.