In eyewitness testimony research, postevent information impairs retention of the original event and increases the probability that interpolated information will be identified as part of the original event. The present experiments studied these effects with 3-month-olds. Infants learned to kick to move a particular crib mobile and then were briefly exposed to information about a novel mobile. The novel postevent information impaired recognition ofthe original mobile when it immediately followed training but not when it was delayed by 1 day. Like adults, infants treated the postevent information as part of the original training event, continuing to do so for at least 2 weeks. We propose that postevent information displaces conflicting information coactive with it in primary memory and creates a new, updated memory token of the event. Once the new token leaves primary memory, however, it is protected; only a copy can be retrieved and modified in the future.