A total of 72 male research participants watched a filmed purse snatching and then assumed the role of an eyewitness and reported the incident. A confederate-interrogator asked the witness to complete a police trait questionnaire describing the perpetrator, or an irrelevant biographical questionnaire. The interrogator then presented a series of mug shots and either emitted subtle, nonverbal cues in an attempt to influence the witness, or did not attempt to influence the witness in any way. Results indicated that neither the questionnaire manipulation nor the bias manipulation affected the accuracy of eyewitness identifications. However, witnesses in the bias condition were likely to identify the target of the interrogator's bias as the perpetrator, even when the biasing cues were directed at nonperpetrators. A total of 14 references are included. (Author abstract modified)