This study tested first whether 240 undergraduate students share high consensual agreement in their selection of faces and voices which fit criminal and non-criminal ‘occupations’ second, whether there are significant relationships between trait impressions and voice characteristics for persons selected as exemplars of criminals and non-criminals; and third, whether categorization of targets into ‘good guys’ and ‘bad guys’ influences recognition memory. Subjects' selections of exemplars of criminals and non-criminals from video recordings of the face and voice, or face-only, or voice-only of 15 white men were done with high confidence in a significantly non-random manner. Significant correlations were found between trait impressions and vocal characteristics in all three presentation modes which differentiated good guys from bad guys. Recognition memory for target voices was significantly inferior to the recognition of face and voice targets, and face-only targets. Recognition confidence scores were significantly higher for good guys over bad guys, especially in the voice-only condition. The results were discussed in terms of the probability that stereotyping of faces and voices can influence decision-making in the legal process.