The prognostic value of the CD4+ T cell count is not clearly established for HIV-1 infected patients with an advanced immunosuppression. The aim of this study was to assess the relationship between CD4+ T cell counts and survival in patients with less than 50 CD4+ T cells per mm3 (/mm3). We examined an historical cohort of 97 patients with 2 consecutive CD4+ T cells determinations < 50/mm3 within 3 months, followed at a university hospital of the University of Montreal. The proportion of men was 93% with 74% being homo/bisexual. The means of the 2 CD4+ T cell counts/mm3 were 25 and 25.1 respectively. Median survival after the first CD4+ T cell count < 50 CD4+ T cells/mm3 was 15.2 months. Using the proportional hazard model, the median survival of patients with 2 consecutive CD4+ T cell counts < or = 20/mm3 was 9.3 months compared to 19.2 for those with 20-50 CD4+ T cells/mm3 (P < 0.0001). It seems then, that the CD4+ T cell count is a helpful prognostic marker, even in very immunosuppressed patients. Its prognostic value is more accurate if the measurement is replaced within 1-3 months because of high variability at this level of immunosuppression.