In clinical trials measuring the short term survival of patients, prognostic stratification is important to avoid susceptibility bias. Demographic and disease specific clinical features are commonly used to established similarity, but these may not insure prognostic comparability. To identify the generic predictors of one year survival after hospitalization, a cohort of 559 patients admitted to the medical service at New York Hospital was studied. Among all of the clinical and demographic information available at the time of admission, only three were independent predictors of the time until death: functional ability, severity of illness and extent of comorbid disease. The overall one year survival rate was 66%. Six subgroups composed of patients with different combinations of the predictors of prognosis had mortality rates ranging from 97% to 14%. Illness severity was a predictor only in the less functional patients, and only severe comorbidity was associated with a decreased survival after taking into account functional ability and illness severity. Prospective estimates of severity of illness, functional ability and extent of comorbidity can be used to stratify a diverse group of patients according to their probability of one year survival after hospitalization.