Psychopathy has proven to be an important construct in the evaluation of mentally disordered offenders, especially for risk assessment. Given the ominous outcomes associated with psychopathy, understanding its underlying dimensions is critical to both construct validation and its forensic applications. This paper begins with an examination of both the traditional two-factor Psychopathy Checklist-Revised (PCL-R) model, which is composed of core personality features and an antisocial lifestyle and the recently proposed three-factor model, which divides the core personality features into two dimensions and eliminates antisocial behavior. A comparative study was then conducted of the traditional two-factor model of psychopathy, the hierarchical three-factor model, and a new four-factor model in a sample of mentally disordered offenders. The study tested the underlying dimensions of the PCL-R. Results are summarized under four main points: (1) the traditional factor structure of the PCL-R does not adequately fit the data in the current sample of mentally disordered offenders; (2) both two and three factor testlet models of the PCL-R fit the data well; (3) the increased focus on personality traits was congruent with several developmental models; and (4) a four-factor PCL-R model integrating the 2001 personality-focused factors with the addition of an antisocial factor produced an excellent fit and an innovative development with the PCL-R. Tables, references