In the authors' earlier study (Marshall et al., 1993), they found such deficits in community-based child molesters but not in incarcerated child molesters. The current study used two scales designed to measure the cognitive (Hogan's Empathy Scale) and emotional (Questionnaire Measure of Emotional Empathy) components of empathy more precisely than measures in the earlier study. Twenty-nine nonfamilial child molesters about to enter treatment at Bath Institution (a medium-security Canadian penitentiary) volunteered to participate in the study. A sample of 29 community males were recruited from a government employment agency by advertising the study in the employment office. In addition to the cognitive and emotional measures, the Personality Questionnaire (Jackson, 1984) was used to measure the tendency of a respondent to portray him/herself in a socially desirable manner. As expected from the authors' predictions, incarcerated child molesters were found to be deficient in the cognitive component of generalized empathy but, contrary to expectations, they were also shown to be deficient in the emotional component. Study results, along with the earlier findings with community- based offenders, provide support for the authors' general theory of child molesting, which suggests that a lack of empathy is a critical feature that allows these men to abuse children. An obvious implication of these findings is that in training child molesters to be more empathic, therapists should focus on broader empathy deficits as well as victim-specific empathy. 1 table and 14 references