Admissable game-theory models of evolution must be restricted to the class of "existential games" in which there is no way of using the winnings ("payoff") for any purpose other than continuing the game for as long as possible. The optimal strategy in such a game is to minimize the stakes played. Organisms and populations seem to be organized consistently with this strategy. The evolution of this organization does not involve group selection or any other deviation from accepted evolutionary theory. The implication of analyzing evolution as an existential game is to deny the idea of evolutionary momentum and to emphasize that adaptedness must always be defined in an explicit environmental context. It is asserted that adaptive responses will have a particular temporal relation to environmental perturbations. This assertion is a predictive statement about evolution-unlike the retrospective assertions that usually emerge from evolutionary analyses.