Some patients taking anti-HIV drugs as part of highly active antiretroviral therapy (HAART) do not show any HIV in the blood; however, some HIV will remain hidden in the body and, if the drugs are stopped, will return to the blood. The purpose of this study is to determine if short periods of stopping HAART increase the activity of CD8 and CD4 cells (cells of the immune system that fight infection), if repeated stopping of these drugs for longer periods of time and restarting them will increase effectiveness of HAART, and if the increased immune system activity as a result of stopping treatment leads to lower levels of HIV over time.

Some HIV infected patients taking HAART have been able to achieve prolonged suppression of HIV viral load for extended periods of time. However, discontinuing HAART has consistently resulted in HIV's return to plasma. Both CD8 and CD4 cells are markedly reduced in individuals with prolonged HIV suppression; control of and response to cell-associated HIV is dependent on immune-mediated mechanisms involving these cells. It is hypothesized that a brief and low-level increase in HIV levels resulting from HAART interruption might boost HIV-specific CD8 and CD4 T-cell counts. After suppression of viral load with the reintroduction of HAART, the expanded CD8 population might be able to better control viral replication and better respond to cell-associated HIV. Future treatment interruption may lead to longer periods of undetectable viral loads.

Patients are divided into 2 age cohorts, with Cohort 1 consisting of children and adolescents 4 years and older up to 21 years of age, and Cohort 2 consisting of children and adolescents 2 years and older up to 4 years of age. Patients will be assigned to one of 2 groups. Group A patients will participate in drug holiday cycles from HAART and then back to HAART; Group B is a control group that remains on continuous HAART throughout the study. Cycle 1 for Group A patients begins with 18 days of HAART and a 3-day drug holiday. At the end of the drug holiday, viral load is measured and HAART is resumed for 28 days (detectable virus cycle) if viral load is detectable after the drug holiday. If viral load remains below the level of detection, the patient begins the next drug holiday cycle. With each subsequent drug holiday cycle, time off HAART will increase by 2 days. Patients failing 4 repeated detectable virus (28-day treatment) cycles will be taken off study.

Patients will be enrolled in this study for a minimum of 142 weeks. For Group A, HIV viral load and CD4 cell count are measured at the end of each drug holiday and each HAART resumption; HIV-specific CD4 and CD8 responses are measured every 3 cycles; and cell-associated HIV is assessed at entry, at 12-week intervals, and at the end of the study. For Group B, physical exams are conducted and HIV viral load and other blood work are measured every 12 weeks.

Note: A Study Monitoring Committee (SMC) recommended on 06/22/05 that this study close, because the primary objectives in this proof-of-concept study could not be achieved. As of 06/28/05, Group A participants will discontinue their drug holiday cycles and will be given the best possible therapy. All participants in this study will have a final study visit to be scheduled as soon as possible, prior to 07/29/05.

Inclusion Criteria:

• HIV infected
• For Cohort 1, CD4 T-cell percent greater than 20
• For Cohort 2, CD4 T-cell percent greater than 25
• Viral load less than 400 copies/ml in the year prior to study entry and less than 50 copies/ml at screening
• Taking anti-HIV drugs (including at least 1 protease inhibitor) and have been on anti-HIV drugs for at least 1 year prior to study entry
• Have been on their current drug regimen for at least 4 months
• Willing to follow study procedures
• Parental or guardian consent if under 18 years old
• Acceptable forms of contraception

Exclusion Criteria:

• Taking abacavir, nevirapine, efavirenz, or delavirdine
• AIDS-related or other infections needing drug treatment at study entry
• Pregnant or breastfeeding
• Have, or have had in the past, diseases (other than HIV infection) or other conditions that, in the doctor's opinion, would interfere with the study
• Taking experimental drugs without the consent of the protocol team