ObjectiveTo assess the safety and efficacy of three new drugs in patients with antiretroviral failure and to correlate retrospectively baseline factors with virological response.Design and settingOpen-label, 48-week, single-arm, multi-center phase II trial conducted at nine US university or government clinics and private practices.Patients and interventionsPatients with HIV-1 RNA ⩾ 500 copies/ml despite ⩾ 20 weeks of treatment with at least one protease inhibitor received abacavir 300 mg twice a day, amprenavir 1200 mg twice a day and efavirenz 600 mg once a day. Other antiretrovirals were prohibited until week 16 except for substitutions for possible abacavir hypersensitivity.Main outcome measuresHIV RNA at weeks 16 and 48.ResultsA total of 101 highly treatment-experienced patients enrolled; 60 were naive to non-nucleoside analog reverse transcriptase inhibitors (NNRTI). HIV RNA < 400 copies/ml was attained in 25 out of 101 (25%) patients at 16 weeks (35% of NNRTI-naive and 10% of -experienced patients) and 23 (23%) patients at 48 weeks (33% of naive and 7% of experienced patients). CD4 cells increased by a median of 15 × 106and 43 × 106cells/l at weeks 16 and 48, respectively. Drug-related rash occurred in 50 out of 99 (51%) of patients, and 17 out of 99 (17%) permanently discontinued one or more drugs as a result. Lower baseline viral load, fewer NNRTI-related mutations, absence of decreased abacavir (⩾ 4-fold) and efavirenz (⩾ 10-fold) susceptibility, and greater number of drugs to which virus was susceptible were associated with virological response at week 16.ConclusionsAbacavir, amprenavir and efavirenz durably reduced HIV RNA and increased CD4 cell counts in a subset of treatment-experienced adults. Baseline viral load and some genotypic and phenotypic markers of resistance correlated with HIV RNA response.