We report the 1st example of a true complementarity effect in memory development—a situation in which memory for the same event simultaneously becomes more and less accurate between early childhood and adulthood. We investigated this paradoxical effect because fuzzy-trace theory predicts that it can occur in paradigms that produce developmental reversals in false memory, which are circumstances in which adults are more likely than children to remember new events as old. The complementarity prediction is this: If subjects separately judge whether those same events are new but similar to old ones, adults will be more accurate than children, even though adults are less accurate when they judge whether the items are old. We report 4 experiments in which children (6- and 10-year-olds), adolescents (14-year-olds), and adults encoded the modal developmental reversal materials: Deese−Roediger−McDermott lists. Then, they responded to memory tests on which half the subjects judged whether test items were old and half judged whether the same items were new−similar. The paradoxical complementarity effect was detected in all experiments: The tendency to falsely remember new−similar items as being old increased with development, but so did the tendency to correctly remember them as being new−similar.