The selection arena hypothesis offers one answer to a puzzling question. Why do some organisms produce many more fertilized zygotes than are actually reared to hatching, birth, or release-then neglect, discard, resorb, or eat some of them, or allow them to eat each other? It makes four assumptions: (1) zygotes are cheap; (2) after conception the investment of parental time, energy, or risk into offspring continues; (3) offspring vary in fitness; (4) variation in offspring fitness can be identified by the mother at an early stage of the life cycle. If these assumptions hold, then one general prediction follows: the parent should overproduce zygotes, identify those with lower expected fitness, then either kill and reabsorb them, let them be eaten by sibs, or simply stop feeding them in order to invest in more promising offspring. The explanation appears to apply to a wide range of phenomena whose common cause had not previously been appreciated. These include: (1) polyovulation in some bats, tenrecs, the plains viscacha, and the pronghorn antelope; (2) cases of recurrent, consecutive, spontaneous abortions in humans; (3) some cases of surplus flower production and fruit abortion; (4) sex-ratio adjustment in red deer, mice, and coypus; (5) some types of cannibalism, including possible cases in mice, sharks, and wasps. Some cases that might be explained by the selection arena hypothesis are also plausibly explained by other causes, including bet-hedging reproductive investment in the face of unpredictable food supplies, and inter-specific or inter-familial aggression as an alternative to parent-offspring or sib-sib cannibalism.