This study examined when eyewitnesses to crime receive feedback about their choice of a suspect from a line-up (or post-identification feedback), such information can substantially alter their recollections of the witnessing experience. This study examined whether feedback exerts similar effects on investigators' recollections of a suspect's behavior. Participant-investigators received training on speech cues that they were told, when present in a speaker's account, signal either honesty or deception. After hearing a suspect's account of a theft, participants decided whether the suspect was lying or telling the truth. One-third of participants subsequently received immediate confirming feedback about their performance, while another third received disconfirming feedback. The remaining one-third of participants did not receive feedback about their decision. Finally, participants rated the frequencies of speech cues that they had been instructed to detect Disconfirming feedback significantly altered retrospective judgments about the characteristics of the suspect's account. Specifically, when told that the decision they made about the speaker's credibility was incorrect, participants judged the speaker as having exhibited fewer behaviors consistent with the credibility decision they had made, relative to those who either received no feedback or confirming feedback. Biases in recollections of a suspect may have consequences in real-world interrogations wherein investigators assess credibility on the basis of numerous behavioral cues. Results are discussed in light of findings of post-identification feedback studies on eyewitnesses. Abstract published by arrangement with John Wiley & Sons.