This analysis examines the predictability of the course of schizophrenia using long-term follow-up data on hospital episodes in a cohort of patients from a psychiatric case register in Denmark. We focus on whether clinical and sociodemographic data collected during the first episode are related to the number of hospitalizations during follow-up and the association of patients' course of hospitalizations with the risk of being rehospitalized. A Poisson regression model and a proportional hazards model were used to address these questions. Age of onset and time to the first rehospitalization were strong early predictors of chronicity of course, as measured by the number of psychiatric hospitalizations for each schizophrenic patient. The results also show that the risk of rehospitalization depends on the previous tenures in the community.