OBJECTIVE: Clinicians' assessments of patients' suicide risk at admission to and discharge from a psychiatric hospital were examined to learn how clinical estimates of risk changed over the course of hospitalization and to identify which demographic and clinical characteristics were associated with higher estimates of risk at admission and discharge. METHODS: Seventy-one treating physicians evaluated risk of self-harm of 241 patients at admission to and discharge from a short-term inpatient unit. Risk within the next week (short-term risk) and within the next year (long-term risk) was estimated. At discharge and admission, the physicians also rated patients' symptoms using the Brief Psychiatric Rating Scale. Nurses rated self-directed aggression during hospitalization with the Overt Aggression Scale. RESULTS: Ratings of short-term risk were lower at discharge than at admission, whereas ratings of long-term risk showed relatively little change. At both discharge and admission, the estimated risk of self-harm was associated with a history of suicidal behavior and with acute symptoms, such as depression, anxiety, and emotional withdrawal. At discharge, the estimated risk was also associated with substance abuse, severity of psychosocial stressors, and living alone. CONCLUSIONS: Clinicians appeared to view their hospital-based interventions as influencing variables relevant to short- term risk of suicide but as having little impact on long-term risk. Implications are discussed for management of suicide risk and for medicolegal assertions regarding prevailing community practices that are made in litigation alleging negligent release of patients from hospitals.