BACKGROUND: Combination antiretroviral therapy administered to HIV-infected individuals has been shown to decrease viral replication, improve immunologic function and delay the progression of HIV infection. However, because patient adherence to complicated combination-therapy antiretroviral regimens is difficult and because of concerns regarding the cumulative toxicity of antiretroviral drugs, regimens that utilize fewer antiretroviral agents are desirable. OBJECTIVES: To compare the use three- or four- versus two-drug antiretroviral maintenance regimens following successful initial therapy for HIV infection. SEARCH STRATEGY: The following electronic databases were searched for relevant randomized trials or reviews: 1. MEDLINE for the years 1982-May 2003 using the search terms human immunodeficiency virus, antiretroviral therapy, maintenance therapy, zidovudine, lamivudine, indinavir, stavudine, saquinivir, nelfinavir, didanosine, zalcitabine, ritonovir, AIDS, anti-HIV agents, HIV infection and HIV seropositivity. 2. AIDSLINE for the years 1982- May 2003 using the search terms antiretroviral therapy, maintenance therapy, zidovudine, lamivudine, indinavir, stavudine, saquinivir, nelfinavir, didanosine, zalcitabine, ritonovir, anti-HIV agents. 3. The Cochrane Database of Systematic Reviews, Database of Abstracts of Reviews of Effectiveness and the Cochrane Clinical Trials Register in the Cochrane Library, through May 2003. 4. AIDSTRIALS, a specialist registry of current and completed trials maintained by the U.S. National Library of Medicine through May 2003. The abstracts of relevant conferences, including the International Conferences on AIDS, the Conference on Retroviruses and Opportunistic Infections, the Infectious Disease Society of America annual meeting and the Interscience Conference on Antimicrobial Agents and Chemotherapy, as indexed by AIDSLINE, were also reviewed. Reference lists of all review articles and primary articles identified were also searched. SELECTION CRITERIA: Randomized controlled trials in which HIV-infected adults who had successfully completed initial three- or four-drug antiretroviral therapy were randomized to maintenance therapy with three or four drugs or maintenance therapy with two drugs. Successful initial therapy was defined by a plasma viral load of less than 500 copies/ml. DATA COLLECTION AND ANALYSIS: Two reviewers assessed eligibility and trial quality. Attempts were made to contact the authors of the included abstract. Data on the number of patients experiencing loss of viral suppression were abstracted by two reviewers. The data were pooled, where appropriate, to yield odds ratios, using random effects models. MAIN RESULTS: Four trials were identified including three published studies and one abstract. Compared to three- or four-drug maintenance therapy, maintenance therapies including fewer drugs were associated with a higher risk of virologic failure (loss of HIV suppression to non-detectable levels). Combining the results of all four studies yielded an odds ratio of 5.55 (95% confidence interval, 3.14 - 9.80). Similar results were obtained when the one abstract was excluded (odds ratio, 5.48; 95% confidence interval, 2.82 - 10.65). Performing subgroup analyses of studies using similar induction and maintenance regimens gave similar results. Maintenance regimens of zidovudine and lamivudine compared to maintenance regimens with zidovudine, lamivudine and indinavir, were associated with significantly higher rates of virologic failure (odds ratio, 4.57; 95% confidence interval, 1.80 - 11.58). Similarly, maintenance regimens that discontinued one or more protease inhibitor after including them in induction therapy were also associated with a significantly higher risk of virologic failure (odds ratio, 6.15; 95% confidence interval, 3.40 -11.10). REVIEWER'S CONCLUSIONS: Although it is desirable to reduce the number of antiretroviral drugs given in combination therapy for reasons of compliance and toxicity, maintenance regimens with fewer drugs are associated with significantly increased resistance and risk of loss of viral suppression. Successful initial therapy, as evidenced by suppression of viral load, should not be modified in the maintenance phase unless clinically necessary.