OBJECTIVE: The purpose of the study was to identify risk factors and correlates of violence committed by patients in an acute adult psychiatric inpatient unit in a district general hospital of the United Kingdom's National Health Service. METHODS: Incidents of violence committed by inpatients over a one-year period in 1997-1998 were retrospectively analyzed. The clinical characteristics of 49 violent patients were compared with those of all patients admitted to the unit during the study period (N=474) and with a random sample of nonviolent patients (N=140). Logistic regression analysis was used to identify clinical variables that predicted violent behavior. RESULTS: Violence was not positively associated with schizophrenia or negatively associated with depression. Frequent medication change, high use of sedative drugs, past violent behavior, an ICD-10 diagnosis of dissocial personality disorder or emotionally unstable personality disorder (DSM-IV antisocial personality disorder or borderline personality disorder), and long hospitalization were the most powerful predictors of violence. Together these variables had a sensitivity of 76 percent, a specificity of 97 percent, and a positive predictive value of 90 percent in predicting which patients became violent. Compulsory (involuntary) admission, comorbid diagnoses, past self-harm, and nonalcohol drug abuse were also associated with violent behavior. CONCLUSIONS: Clinicians' judgment about an inpatient's potential for violence may be augmented by knowledge of the risk factors identified in this study. Medication variables could be especially useful predictors, particularly when information about other risk factors is not available. Factors other than mental illness per se may be crucial determinants of violence in acute inpatient settings.