The authors compared 2-, 3-, 4-factor, and 2-factor/4-facet Psychopathy Checklist-Revised (PCL-R) models in a previously unpublished sample of 1,566 adult male sex offenders assessed under applied clinical conditions as part of a comprehensive state-mandated community notification risk assessment procedure. "Testlets" significantly improved the performance of all models. The 3-factor model provided the best fit to the current data, followed by the 2-factor/4-facet model. The 2-factor model was not supported. (Contains 3 notes, 4 tables, and 1 figure.)