The control of cheating is important for understanding major transitions in evolution, from the simplest genes to the most complex societies. Cooperative systems can be ruined if cheaters that lower group productivity are able to spread. Kin-selection theory predicts that high genetic relatedness can limit cheating, because separation of cheaters and cooperators limits opportunities to cheat and promotes selection against low-fitness groups of cheaters. Here, we confirm this prediction for the social amoeba Dictyostelium discoideum; relatedness in natural wild groups is so high that socially destructive cheaters should not spread. We illustrate in the laboratory how high relatedness can control a mutant that would destroy cooperation at low relatedness. Finally, we demonstrate that, as predicted, mutant cheaters do not normally harm cooperation in a natural population. Our findings show how altruism is preserved from the disruptive effects of such mutant cheaters and how exceptionally high relatedness among cells is important in promoting the cooperation that underlies multicellular development.ooperation is a hallmark of major transitions in biological complexity: from molecules to genes, from genes to chromosomes, from primitive cells to complex cells, from cells to multicellular organisms, and from multicellular organisms to societies (1-3). Cooperative groups are vulnerable, however, to exploitation by cheaters, individuals that have access to group benefits without contributing their fair share (1-3). Among cells and individuals, high relatedness is thought to aid in selection against cheaters (4-6). High relatedness means that cheaters and cooperators will tend to be in different groups, which both limits opportunities for cheaters to exploit cooperators and exposes any group-level defects of cheaters to selection. Curiously, although such control is central to selfish-gene theory, tests at the genetic level have been limited by the kinds of information available. In large organisms, relatedness is often estimated, but cheater genes are unknown. In microorganisms, cheater genes can be found (7-13), but little is known about relatedness in natural social groups.The life cycle of social amoebae presents a challenge to the importance of relatedness in promoting selection against cheaters and an opportunity to test it. When the normally solitary amoebae are starved of their bacterial food source, they gather into a multicellular aggregate that forms a fruiting body. Here, Ϸ25% of cells altruistically die, forming a stalk that holds up the remaining cells, differentiated as spores, for dispersal (14-17). Thus, unlike more familiar organisms that develop from one cell, development begins by aggregation of many dispersed cells. Different clones can mix and cheat each other (18,19), for example by avoiding contributing to the sterile stalk (7). Models (20)(21)(22), experiments (7,23,24), and a natural observation (24), suggest that cooperative fruiting body formation can be threatened by the spread of mutant cheat...