This study explores suicide rates as a quality measure and identifies risk factors for suicide among psychiatric inpatients. Data from a prospective mortality study of psychiatric inpatients from 128 U.S. Department of Veterans Affairs hospitals throughout the United States were utilized to examine the relationship of death by suicide to patient-level sociodemographic, clinical, and mental health service delivery characteristics, as well as to facility-level measures of service delivery. Data were collected on all patients discharged with a diagnosis of schizophrenia, major depression, posttraumatic stress disorder, or bipolar disorder (N=121,933) between 1994 and 1998. There were 481 suicides in the study sample. Several variables were associated with higher risk for suicide risk, including length of stay less than 14 days, poorer continuity of care, and lack of readmission within 6 months. These variables were significant even after adjustment for state suicide rates, the proportion of members of racial minority groups in the state population, and state-level social capital. Suicide rates are not likely to be a useful indicator of the quality of mental health care because of the instability of suicide rates, difficulty collecting data, and the lack of association between suicide and facility quality of care. The lack of association with facility-level variables suggests that systemic changes in these performance measures would be unlikely to significantly reduce suicide rates.