Based on the Yudofsky scale, a Modified Overt Aggression Scale (MOAS) with upgraded psychometric properties was developed to assess the nature and prevalence of aggression in a psychiatric population. The present report describes the standardization of this scale and the pattern of findings on two cohorts of 114 and 150 inpatients. The results support the discriminative validity of the MOAS and its internal, interrater, and retest reliabilities. Within 1 week some form of aggression was noted in about one fourth of the patient samples, with verbal aggression the most prevalent and autoaggression the least. Chronic patients showed the lowest incidence of physical assault and general aggression, whereas gender differences and daily variations were not significant. Greater stability of aggression was demonstrated for patient groups and for forms of aggression with higher base rates and for the short term (within 1 week) rather than the long course (3 months). The high prevalence of aggression and the consistency of profiles across patient samples suggested that sensitive, multivariable scaling can improve the accuracy of measurement and the depiction of the construct.