We ran an experiment with children to study the development of honesty with age. We asked each child to toss a fair coin in private and to record the outcome (white or black) in a paper sheet. We rewarded only those who reported white. We found a fraction of reported whites significantly larger than 50%, uniformly across age groups. This suggests that some children cheat when cheating is profitable and they are not observed. In a second treatment we told children not to cheat. This reminder reduced the probability of reporting white by 18% on average, and significantly more in girls.