Group size has long been considered to affect the benefits of group-living. Certain group sizes have better predator detection rates, access to higher quality territories, and higher reproductive success. Several years ago we documented the occurrence of kidnapping in our study species (the Pied Babbler), a rare event that we hypothesized was due to the benefits of group size. We predicted that groups would benefit from kidnapping other groups’ young rather than face the risk of extinction during years when they failed to raise their own young. Following the accumulation of twelve years of long-term data, we have formalized our predictions theoretically and empirically. We have found that at very small group sizes, kidnapping young confers a selective advantage. Our empirical data confirmed model predictions: smaller groups were more likely to go extinct than larger groups, and kidnapping occurred primarily in groups that had failed to raise their own young. By showing that individuals invest in kidnapping (a costly behaviour) in a predictable manner, we conclude that this behaviour is related to the benefits of maintaining group size, thus supporting the hypothesis that there is a selective benefit of group size that may promote the evolution of social living.