Context: In-hospital and postdischarge survival rates of geriatric hospitalized patients are poorly documented. Objective: Assess and compare the characteristics and long-term survival of cohorts of old hospitalized patients over three decades in the same university geriatric structure. Compare the survival rates with the survival of the general population. Design: Retrospective observational study of the first admission of patients admitted in 1976, 1986 and 1996. Setting: University geriatric teaching hospital of 300 beds located in Geneva, Switzerland. Survival data was available in 66.6, 96.5 and 99.9% of the 1206, 1581 and 1951 patients discharged alive in 1976, 1986 respectively 1996, as well as demographic characteristics (age, gender, marital status) and other medical conditions (medical diagnoses, number of admissions and length of stay). The survival rates of the Geneva general population have been standardized for age, gender and period. Results: The mean age of patients increased significantly over time (p <.0001): 80 years in 1976, 83 years in 1986 and 84 years in 1996. The oldest patient was 98.1, 102 and 102.2 years old, respectively. The number of diagnoses was similar in the three cohorts with a median of four per patient. The annual number of hospitalizations increased over time, with a median survival after hospital discharge of 7.5 years for the 1976 cohort, 3.8 years for the 1986 cohort and 3.2 years for the 1996 cohort. The post discharge survival was significantly higher for the 1976 cohort than the two others. The 6-year survival rates of the 1976, 1986 and 1996 cohorts, respectively 64, 35 and 29% were significantly different (P <.0001). Conclusions: The analysis of the 30-year experience revealed significant changes in the characteristics of hospitalized patients. Survival length after a geriatrics hospitalization is substantial.