3–5-year-olds heard a story involving identical twins, one of whom was absent when their ball was moved from one drawer to another. Children found it easy to infer that the twin who later went to the original location to get the ball was the one who had gone outside. Children in a comparison condition found it relatively difficult to predict where a (nonidentical) twin who was absent when the ball was moved, would search for the ball, and made the usual realist error. In further investigations involving variations on the identical twins task, children were equally successful at making the link between looking in the wrong place and having been absent, whether a backwards inference was required (as above) or a forwards one (inferring that the twin who went outside must now be the one who was at the wrong location). We ruled out one twin's physical association with the correct location as an artifactual explanation for facilitation. Children performed well whether or not the experimenter told them explicitly which twin did not know the ball had been moved. These findings support the view that children's early insight into the representational character of mind is masked in traditional prediction tests of false belief.