BACKGROUND: Structured treatment interruption was evaluated in 74 patients who had been pretreated with antiretrovirals, consisting of 2 nucleoside reverse-transcriptase inhibitors (NRTIs) for 1 year followed by 3 years of highly active antiretroviral therapy containing a protease inhibitor. METHODS: Patients with a CD4 cell count of > or =350 cells/microL and a plasma viral load of <50 copies/mL were randomized to 3 therapy arms: (1) continuous therapy, (2) CD4 cell count-guided theory, and (3) week-on/week-off (WOWO) therapy. The efficacy and safety of structured treatment interruption and antiretroviral use were evaluated in human immunodeficiency type 1 (HIV-1)-infected patients. The study end points were percentage of patients who developed AIDS or who died and a CD4 cell count of > or =350 cells/microL. Intergroup differences were analyzed using analysis of variance and Kruskal-Wallis tests. RESULTS: Baseline characteristics at the start of the structured treatment interruption were similar. At week 48, no patient had died, and 1 patient in the WOWO group had an AIDS-defining condition. The proportions of patients with a CD4 cell count of > or =350 cells/microL were 100%, 87%, and 96% in treatment arms 1, 2, and 3, respectively. The percentages of weeks of antiretroviral use were 100%, 41.1%, and 69.8% in arms 1, 2, and 3, respectively. The adverse events were not significantly different among arms (P=.27). Thirty-one percent of patients in the WOWO group experienced virological failure. CONCLUSION: WOWO therapy maintained a CD4 cell count of > or =350 cells/microL in almost all patients but was associated with high virological failures rates (possibly resulting from previous dual-NRTI therapy), indicating that this strategy is less useful. Receipt of CD4 cell count-guided therapy resulted in comparable clinical outcomes to continuous therapy and may save antiretroviral-associated costs, but this needs to be confirmed by a larger trial.