OBJECTIVE: To assess the safety of 2 intermittent treatment strategies compared with continuous therapy for patients with virologic suppression on highly active antiretroviral therapy (HAART) at baseline. DESIGN: Seventy-four nucleoside reverse transcriptase inhibitor (NRTI) and protease inhibitor (PI) pretreated patients with an HIV RNA level <50 copies at screening were randomized to continuous treatment, CD4-guided treatment, or week-on-week-off treatment with 2 NRTIs plus 1600 mg/100 mg of saquinavir/ritonavir once daily. At week 96 (end of the randomized phase of the study), all patients were given continuous HAART for 12 weeks to week 108. Primary outcomes were the proportion of patients with a CD4 count >350 cells/microL and HIV RNA level <400 copies/mL at week 108. METHODS: Patients were followed up every 12 weeks for CD4 count, HIV RNA level, and clinical and laboratory toxicities. In the CD4-guided arm, treatment was stopped and restarted using a CD4 count threshold (above or below 350 cells/microL or reduction of 30%). RESULTS: Seventy-four patients were enrolled with a median CD4 count of 644 cells/microL before the structured treatment interruption (STI). The week-on-week-off arm (n=26) was discontinued at week 72 because of high rates (46%) of HIV RNA rebound above 50 copies/mL. In the continuous arm, 25 (100%) of 25 patients and 24 (96%) of 25 patients had an HIV RNA level <400 copies/mL and <50 copies/mL, respectively, at week 108, and 96% had a CD4 count above 350 cells/microL, with a median CD4 count of 661 cells/microL. Patients in the CD4-guided arm had a significantly lower median CD4 count (489 cells/microL) than the patients in the continuous arm (P=0.03), but all had a CD4 count above 350 cells/microL and 1 had a new HIV-related illness. At week 108, 21 (91%) of 23 patients and 13 (57%) of 23 patients had an HIV RNA level <400 copies/mL and <50 copies/mL, respectively. Those who did not achieve an HIV RNA level <50 copies/mL had a higher HIV RNA load before retreatment, and 4 of 5 patients subsequently achieved viral suppression after an additional 12 weeks of HAART (week 120). Therefore, 17 (94%) of 18 evaluable CD4-guided arm patients achieved viral suppression after retreatment. Antiretroviral (ARV) side effects were similar in all arms. CD4-guided treatment had a 54% ARV cost savings. CONCLUSIONS: This pilot study suggests that CD4-guided HAART is a well-tolerated and cost-saving treatment strategy for patients with high pre-ARV and pre-STI CD4 counts. Week-on-week-off treatment had a high virologic failure rate and was discontinued. The HIV RNA suppression rate was similar in patients treated with continuous HAART and in those retreated with 12 to 24 weeks of HAART after CD4-guided therapy.