After a staged interruption in a college classroom, 57 participants in the first experiment were questioned about the event five times over 5 weeks and 79 participants in the second experiment were questioned about the event three times over 5 days. During the final questioning session, participant witnesses consistently reported higher levels of confidence for items that had been subject to repeated postevent questioning than for items that were asked for the first time. Yet, no differences were observed in the accuracy of responses to the two sets of items. Additionally, in all conditions, participant witnesses were generally overconfident in their responses. Results suggested that repeated postevent questioning may cause confidence estimates of eyewitnesses to be artificially inflated. Theoretical and practical implications of the findings are examined, and recommendations for the legal system are offered. 59 references and 5 figures