To investigate the clinical, virologic and immunologic consequences of planned treatment interruptions in chronically HIV-infected patients. One hundred forty-one patients with undetectable viral load for at least 6 months and CD4+ T cells count greater than 500 per microliter were recruited. Their antiretroviral therapy was stopped and clinical, analytic, virologic, and immunologic data were recorded at baseline, during discontinuation, and after restarting treatment. Viral load rebound after discontinuation in 137 (97%) patients, and was similar to prehighly active antiretroviral therapy (HAART) levels. A rapid decrease in CD4+ T-cell count (median, 240 cells per microliter), was observed in the first 3 months in all patients, with pronounced differences between them. After a median follow-up of 36 months, 45.5% patients were still without therapy. Factors related to a more severe decline were a prior lower CD4+ T nadir (<200 cells per microliter) before starting HAART, a greater increase (>500 cells per microliter) with it, a higher CD4+ T-cell count before interruption (>800 cells per microliter) and a higher viral load rebound after it. The increase in CD4+ T-cell counts after reinitiation was slower than the decline and only 55% of patients have regained the preinterruption levels at 12 months of follow- up. Twelve infectious events were registered. Treatment failure related to drug resistance was observed in two patients. Planned treatment interruptions may be safe in selected patients with previous CD4+ T cell nadir greater than 200 cells per microliter and pre-HAART VL less than 55.000 copies per milliliter, but should be not recommended in patients with the prognostic factors related to a rapid decline described in this study. Furthermore, there is a considerable concern about the development of drug resistance and the possibility of an incomplete immune reconstitution after the treatment interruption in some patients.