Natural speech contains a wealth of information relevant to understanding cognitive and affective psychological processes, which are reflected in both prosodic and semantic channels of communication. While differences in semantic channels have been demonstrated among psychopathic versus non-psychopathic individuals, research on the role of prosody in psychopathy is scant. The Computerized Assessment of Natural Speech protocol provides adetailed assessment of macroscopic-level prosody variables related to underlying psychological processes that have been linked to psychopathological conditions. Psychopathy is a condition that involves a number of disruptions in cognitive and affective processes, which theoretically can be tied to various aspects of speech. The present study provides a novel contribution by examining natural speech output in an offender sample in the context of a clinical interview (Psychopathy Checklist – Revised). More specifically, the present study examined variance in prosody across segments of the PCL-R interview designed to elicit both positively and negatively valenced emotional content, across high and low levels of subjective arousal, in psychopathic (n = 49) and non-psychopathic (n = 44) male offenders who were similar in terms of age, education, race/ethnicity, and IQ. Three-factor mixed MANOVAs (Group x Valence x Arousal) were conducted to evaluate differences in prosodic speech displayed by the offenders. Results indicated significant interactions between psychopathic and non-psychopathic offenders across valence and arousal conditions in terms of percentage of silence, average pause length, longest pause length, average within-utterance variation in subjectively defined pitch and articulation variables, and average rate of change in articulation across speech sample. Implications and future directions for research are discussed.