How much a person is affected by postidentification feedback is dependent on the credibility of the person giving the feedback. Seven hundred and ninety participants across three experiments viewed a crime video, made judgments from a line-up, were provided with co-witness and/or outcome feedback (from police officers [high credibility] or children [low credibility]), and answered testimony-relevant questions (e.g. How good a view did you get of the person in the video?). The aim was to find out how high versus low credibility co-witness feedback affects a witness' retrospective judgments (Experiment 1) as well as estimations of these co-witnesses' judgments (Experiment 2). Experiment 1 showed that the feedback effect was only observed when the co-witness responses were attributed to a high credibility source. Experiment 2 showed that high credibility co-witnesses were estimated to score higher on the testimony-relevant questions as compared to low credibility co-witnesses. Experiment 3 showed that outcome feedback (e.g. ‘you identified the suspect’) produces stronger effects on testimony-relevant questions than co-witness feedback. The implications of these findings are that when postidentification feedback is present, it is important to determine the source of this feedback. Copyright © 2008 John Wiley & Sons, Ltd.