Measurements of impulsive and premeditated aggression, developed recently, have been applied to prison and outpatient populations without severe mental disorders. Comparable measures of impulsive and premeditated aggression have not been developed for populations with a severe mental disorder. A practical difficulty is that seriously disturbed, thought-disordered patients are incapable of providing reliable historical information. The investigators adapted the Barratt-Stanford instrument for differentiating impulsive from premeditated aggression so that instead of serving as an interview schedule, it could be used to assess aggression from previously documented written descriptions. The study found that the majority of ratable patients showed predominantly impulsive aggression, and after omitting four weak items, the inter-rater reliability for the determination of impulsive aggression was good (k = 0.53). Far fewer of the patients were determined to have shown predominantly premeditated aggression (from 14.2% to 15.5%) and the inter-rater reliability for premeditated aggression was deemed fair (k = 0.33).