Our objective was to assess, in the clinical setting, the predictors of immune reconstitution (IR) and its relation with long-term clinical benefit, in HIV patients with advanced disease after highly active antiretroviral therapy (HAART) through an observational study. A retrospective cohort study in a clinical setting of 383 consecutive adult patients with advanced HIV infection (CD4+ cells <200/mm(3) at baseline), starting their first protease inhibitor (PI)-containing regimen was observed. Immune reconstitution was defined as CD4 count >200 cells/mm(3) and an increase > or =100 cells from baseline, anytime since starting HAART. Clinical benefit was defined as decreased mortality and reduction in AIDS-defining events, AIDS-related complex (ARC) events, major infections and hospitalization (days spent in hospital). During a mean follow-up of 808 days, 261 patients (68.1%) achieved IR. About 50% of these patients reached this result within one year after starting HAART. In multivariate analysis, predictors of immune recovery were sex (female) and baseline CD4 count higher than 50 cells/mm(3). The group of patients with IR had greater clinical benefit with lower mortality, fewer AIDS-defining events, shorter lengths of stay in hospital, fewer ARC events and fewer major infections during all the follow-up (P < 0.0001, tests for trends). However, although they did less remarkably than the first group of patients, even those patients who did not achieve IR experienced a significant decrease in the incidence of all the above events, as compared with the first and sometimes the second trimester after starting their HIV therapy. About 70% of HIV patients with advanced disease achieved IR after starting HAART. Such a benefit is a time-dependent effect and may even take more than 2 years to occur. Predictors of IR were sex (female) and higher baseline CD4 count (>50 cells/mm(3)). The patients who achieved immune recovery performed clinically better than patients who did not. Also the patients who failed to gain such a strong immunological recovery experienced a long-term clinical benefit. This suggests that PI-containing regimens, in advanced HIV disease, may produce a significant clinical benefit, at least temporary, even for patients who do not achieve a substantial immune response.