Hospital-based treatment for HIV patients consumes considerable health care resources, and for planning purposes it is important to know the hospital needs for each stage of infection. In this paper, we report the results of a national study in Italy on hospital stay for cohorts of patients hospitalized for the first time with HIV-related illnesses during the years 1984-86 and 1987-88. Using the Kaplan-Meier survival methods, we demonstrate a substantial decrease in the proportion of time spent in the hospital by patients newly diagnosed with AIDS (from 49% to 25%) in the latter study period. Treatment with zidovudine (ZDV), estimated only for 1988, emphasizes this trend even more. A reduction in hospital use also occurred for AIDS-related complex (ARC) and persistent general lymphadenopathy (PGL) patients. Possible reasons for this decrease in proportion of time spent in the hospital include longer life span of patients and expansion of outpatient care, as has already been reported in the United States.