Abstract The aim of this study was to determine the factors related to outcome, as measured by the cumulative number of days hospitalized, of a cohort group of inpatients with a diagnosis of schizophrenia followed up for 6 years. Comprehensive data were obtained from 18 psychiatric hospitals from two surveys conducted in 1993 and 1999. Outcome was evaluated by calculating the cumulative number of days between the two surveys that the cohort group was hospitalized. Tree-based models analysis was used to explore the factors related to outcome. Patient characteristics at the time of the first survey that were related to outcome were the number of days hospitalized, cumulative number of years hospitalized, age, Global Assessment of Functioning score and relationship to the main visitor. The number of beds in each hospital and changes in their number were also related. Increasing staffing levels is not likely to impact on deinstitutionalization since their levels were not related to outcome.