Abstract. Limited dispersal may favor the evolution of helping behaviors between relatives as it increases their relatedness, and it may inhibit such evolution as it increases local competition between these relatives. Here, we explore one way out of this dilemma: if the helping behavior allows groups to expand in size, then the kin-competition pressure opposing its evolution can be greatly reduced. We explore the effects of two kinds of stochasticity allowing for such deme expansion. First, we study the evolution of helping under environmental stochasticity that may induce complete patch extinction. Helping evolves if it results in a decrease in the probability of extinction or if it enhances the rate of patch recolonization through propagules formed by fission of nonextinct groups. This mode of dispersal is indeed commonly found in social species. Second, we consider the evolution of helping in the presence of demographic stochasticity. When fecundity is below its value maximizing deme size (undersaturation), helping evolves, but under stringent conditions unless positive density dependence (Allee effect) interferes with demographic stochasticity. When fecundity is above its value maximizing deme size (oversaturation), helping may also evolve, but only if it reduces negative density-dependent competition.Key words. Allee effect, altruism, demographic stochasticity, environmental stochasticity, kin selection, recolonization. Behaviors by which an individual sacrifices a fraction of its resources or itself to the benefit of another individual are of central importance to the evolution of sociality. Evaluating the conditions under which such behaviors are selected for requires a careful account of how the behavioral effects translate into fitness costs and benefits. A considerable amount of theoretical work has been devoted to this issue, usually under the heading of ''evolution of altruism.'' Inclusive fitness theory (Hamilton 1964) emphasizes that altruism may evolve if recipient individuals tend to bear the genes underlying the behavior. Thus, genetic relatedness between actor and recipients matters for the evolution of such behaviors. However, it has been more recently emphasized that genetically related neighbors are also more strongly competing for the same local resources, which could at least partially offset the fecundity benefits to neighbors. Indeed, in a simple model where relatedness between group members is induced by population structure, the direction of selection on helping is determined solely by direct fecundity benefits, the behavior being selected for only if the actor's fecundity, that is, the number of juveniles counted before any competition stage, is increased (Taylor 1992a). This result takes the form B/N Ϫ C Ͼ 0, where N is the group size, C is the fecundity cost of the act, and B is the total fecundity benefit to the group, including the actor, so that a benefit B/N is received by the actor.Taylor's model assumes a fixed number of adults. With this assumption, the behavior may affect t...