To compare the clinical, immunologic, and virologic outcomes of efavirenz (EFV)-based versus protease inhibitor (PI)-based highly active antiretroviral therapy (HAART) in severely immunosuppressed HIV-1-infected patients. Retrospective observational cohort study. Responses were analyzed according to the intent-to-treat principle among antiretroviral-naive patients with < 100 CD4 cells/muL who started EFV (n = 92) or a PI (n = 218) plus 2 nucleoside reverse transcriptase inhibitors. The primary end point was time to treatment failure. Secondary end points were percentage of patients with a viral load < 400 copies/mL, time to virologic failure, time to CD4 lymphocyte count > 200 cells/microL, and incidence of opportunistic events or death. The median baseline CD4 cell count and viral load were 34 cells/microL and 5.54 log10 copies/mL (EFV group) and 38 cells/microL and 5.40 log10 copies/mL (PI group). Time to treatment failure was shorter with a PI-based regimen than with an EFV-based regimen (adjusted relative hazard [RH] = 2.19, 95% confidence interval [CI]: 1.23-3.89). After 12 months of therapy, a significantly higher proportion of patients receiving EFV reached a viral load < 400 copies/mL (69.4 vs. 45.1%; P < 0.05). The probability of virologic failure was higher with a PI than with EFV (adjusted HR = 2.52, 95% CI: 1.14-5.61; P = 0.024). There was no difference in time to CD4 cell count > 200 cells/microL or in incidence of opportunistic events or death. : In severely immunosuppressed, antiretroviral-naive, HIV-1-infected patients, treatment with an EFV-based regimen compared with a nonboosted PI-based regimen resulted in a superior virologic response with no difference in immunologic or clinical effectiveness.