For more than a decade, researchers and practitioners have generally accepted a two-factor model for the Psychopathy Checklist-Revised (PCL-R) composed of core personality features and an antisocial lifestyle. Very recently, Cooke and Michie (2001) proposed a three-factor solution that divided the core personality features into two dimensions while eliminating antisocial behavior. This study of male, mentally disordered offenders (N = 96) directly compared factor models via confirmatory factor analysis (CFA). When using testlets to combine theoretically similar items into single ratings, the nested three-factor model was an excellent fit. Of importance, the development of a four-factor model with the inclusion of antisocial items also produced an excellent fit. Combined with recent research, these findings have important ramifications for the construct validity of the PCL-R.