Data from 9,055 adult intakes performed over two and a half years of a managed mental health care demonstration project in a large U.S. city were used as indirect measures of quality of care. The level of care to which patients were initially assigned was examined in relation to the patients' clinical status as judged by both managed care case managers and treatment providers. During the period, assignment to inpatient care of patients in almost every clinical category decreased. The decrease seemed to reflect a policy decision to limit use of all inpatient services rather than a selective elimination of unnecessary hospitalization. Case managers rated a smaller proportion of patients severely disturbed, partial hospitalization was rarely used as an alternative to inpatient care, and detoxification service were increasingly used as an inpatient alternative. Although these data reinforce common beliefs about managed care, the quality of managed care programs can be accurately assessed only with data collected specifically for evaluation purposes.