Psychopathy ratings that employed the PCL-R (Hare, 1985, 1991) were compared in one clinical and three research samples (total N = 285). Differences among the samples were not related to whether ratings were completed under the expressed promise of confidentiality of a research context vs. a pre-parole psychological assessment. The problem of decision errors in prediction is highlighted to address the difficulty in integrating the PCL-R into correctional policy. General issues related to the clinical application of the PCL-R also are discussed.