A rational explanation for cannibalism is that it would be favored under conditions of crowding of conspecific individuals and/or low availability of alternative prey with the fear of starvation, so as to maximize individual fitness. Cannibalism has, however, not evolved and is not maintained by a simple individual optimization, while it has evolved and is maintained as a game among population members. We analysed the attainable state of an evolutionary cannibalism game within a framework that reflects the minimum essence of cause-effect in the cannibalism phenomenon. Cannibalism is predator-prey interaction among conspecifics. Immediate direct payoffs (survival in the interaction among conspecifics) and indirect payoffs (growth results in potential productivity and survival against the threat of starvation) would be included. No morphological specialization and no size priority of cannibalism individuals are assumed as conservative situations in which we analyse the possibility of cannibalism. Cannibalism would be possible under the conservative condition, if initially the wild population's cannibalism rate is not sufficiently lower than a threshold value. Crowding and/or low availability of alternative prey with the fear of starvation facilitates cannibalism evolution. Energy gain from conspecific prey would be attenuated by costs of counterattacks by conspecific victims and by challenge cost of its own. Discounting net intake energy required in the arms race for cannibalism challenge result in a relative disadvantage of having a high cannibalism rate and makes an evolutionary equilibrium of low cannibalism rate, even when potential profitability of conspecific prey is high.