Recent research has provided some clarification about the relation between mental illness and violence with a large body of research on several “state-of-the-art” risk assessment tools considerably improving the ability to identify mentally ill people who are likely to be involved in violence. Risk assessment tools have implications for risk management and intervention with violence prone patients. This study provides a detailed view of the types of civil psychiatric patients reliably identified by the by the multiple Iterative Classification Tree (ICT) as at high risk for violence. The study sought to determine whether hypothesized subtypes of patients existed in a select high-risk sample to promote the future development of targeted treatment programs for these patients. The results indicate that these high-risk patients are best classified into one of three groups: (1) alpha patients (48 percent), who generally are depressed and abuse drugs heavily and manifest multiple core traits of psychopathy and have extensive histories of arrest; (2) beta patients (38 percent), who typically are dysphoric, dependent on alcohol and other drugs, and are sensitive to personal problems; and (3) delta patients (14 percent), who typically are delusional, have less drug and alcohol involvement, often experience command hallucinations, and have histories of intensive treatment. These three groups are largely consistent with hypothesized subtypes. These results provide a framework for systematically investigating the differential effects of particular types of interventions with these patients. Tables, notes, references