A total of 78 rapists and 31 child molesters who were serving sentences of 2 years or more in the Ontario region of the Canadian Federal penitentiary system were eligible for the study. They were recruited during the period from October 1996 to November 1998 from consecutive admissions to the Regional Treatment Centre Sexual Offender Programme for assessment and/or treatment. The rapists were all incarcerated for sexual or sexually related offenses against females 16 years old or older; the child molesters had committed sexual offenses against a child 12 years old or younger. On or shortly after admission, each man was administered a phallometric assessment that consisted of the Quinsey sexual violence assessment and the Barbaree sexual violence assessment. During the testing, penile circumference was measured with a DM Davis mercury-in-rubber strain gauge to assess arousal under exposure to various testing stimuli. Data on historical variables were obtained for each man. Rapists' and child molesters' responses to each of the stimuli were compared via t tests. Significant differences were found in responding to the physical assault-robbery motive stimulus from the Quinsey set and the reluctant participation stimulus from the Barbaree set. In both cases, child molesters had higher average responses than rapists. Rapists were classified into three groups based on peak deviance indexes: men with a clear preference for consensual activity, an ambiguous group, and men with a clear preference for rape. No significant links were found between sexual arousal as determined by phallometric assessments and offense-related variables, psychopathy, or risk-assessment measures. Because the two stimulus sets used in this study failed to agree on the classification of approximately 50 percent of the offenders, such phallometric data cannot be deemed reliable. 5 tables and 37 references