The study evaluated the overall survival after AIDS diagnosis of 1,014 patients reported to the Italian AIDS Registry as resident in Tuscany, stratified by age, gender, year of diagnosis, HIV transmission category, initial AIDS-defining disease and CD4+ cells count. The study was a population-based survival analysis, carried out through Kaplan-Meier method (mean survival times-MST-, 1, 2 and 3-year observed survival) and Cox models (crude and adjusted relative risk-RR). The MST was 12.4 months for all cases, increasing from 4-7 months in 1985-1987 to 14 months in 1991-1992. The observed survival was 51.4% at the first year of follow-up, 28.4% at the second year and 14.5% at the third year. The multivariate analysis showed an independent prognostic effect of age, year of diagnosis, initial AIDS-defining disease and CD4+ cells count. The prognosis was worse in cases aged over 44 (reference: 25-29), diagnosed before 1988 (reference: 1991) and with wasting syndrome, toxoplasmosis, HIV encephalopathy or multiple diseases (reference: PCP alone); and better in cases with more than 100 CD4+ cells/mm3 (reference: < or = 50 cells/mm3). The differences in gender and among HIV transmission categories disappeared after age-adjustment. The study confirmed, in an European population-based series, the poor long-term AIDS prognosis and, once AIDS has became clinically manifest, the prognostic value of some clinical and demographic variables.