Nursing homes have played a major role in deinstitutionalization, and their increased use for the mentally ill has been questioned. We performed a controlled study of nursing homes as an alternative to continued psychiatric hospitalization. Men (N = 403) referred for nursing home placement from eight Veterans Administration medical centers were randomly assigned to community nursing homes (CNHs), Veterans Administration nursing care units, continued care on the same ward, or transfer to another psychiatric ward. Patients met defined criteria for schizophrenia or organic brain disease. Data were collected before random assignment and six and 12 months later, covering physical and mental function, psychopathology, mood, social adjustment, satisfaction with care, as well as drug use, characteristics of settings, and movement in and out of settings. Significant differences between settings were found in self-care, behavioral deterioration, mental confusion, depression, and satisfaction with care. Results were strikingly consistent, showing the group transferred to another ward doing better and the CNH group doing worse. Drug use did not differ from six months before entering the study or later between the settings. Cost showed a marked advantage for the CNH group. Thus, the less costly community nursing home alternative must be viewed in the context of the nonmonetary costs of less favorable patient outcome.