The purpose of this 2-year, case control study was to determine whether differentiation between assaultive and nonassaultive patients can be made based on behavioral assessments and/or sociodemographic variables. For each assault incident, the chart of the patient who assaulted and a randomly chosen patient who did not assault on that day were reviewed (N = 72 subjects or 26 pairs). Various scales were used to evaluate the subjects retrospectively (the day prior to the assault), and patients who assaulted staff were interviewed when possible. An analysis found no differences between patients who assaulted and controls on sociodemographic variables. Those who assaulted had significantly more prior assaults (p = .04) and more difficulty verbalizing angry feelings appropriately on their units (p < .01). Prior to the assault, assaultive patients were more verbally hostile (p = .037) and showed more increased motor activity (p = .001) than controls.