The study of human aggression has been hindered by the lack of reliable and valid diagnostic categories that specifically identify individuals with clinically significant displays of impulsive aggressive behavior. DSM intermittent explosive disorder (IED) ostensibly identifies one such group of individuals. In its current form, IED suffers from significant theoretical and psychometric shortcomings that limit its use in clinical or research settings. This study was designed to develop a revised criteria set for IED and present initial evidence supporting its reliability and validity in a well characterized group of personality disordered subjects. Accordingly, research criteria for IED-Revised (IED-R) were developed. Clinical, phenomenologic, and diagnostic data from 188 personality disordered individuals were reviewed. IED-R diagnoses were assigned using a best-estimate process. The reliability and construct validity of IED-R were examined. IED-R diagnoses had high interrater reliability (kappa = .92). Subjects meeting IED-R criteria had higher scores on dimensional measures of aggression and impulsivity, and had lower global functioning scores than non-IED-R subjects, even when related variables were controlled. IED-R criteria were more sensitive than DSM-IV IED criteria in identifying subjects with significant impulsive-aggressive behavior by a factor of four. We conclude that in personality disordered subjects, IED-R criteria can be reliably applied and appear to have sufficient validity to warrant further evaluation in field trials and in phenomenologic, epidemiologic, biologic, and treatment-outcome research.