This paper explores the question of when (or if) a market leader firm is best off with a strategy of product cannibalism: introducing a new product designed to supersede and hence destroy its own current bestseller before a rival does. Particular attention is given to the payoffs of various superseding product strategies and, given these strategies, whether the leading firm can be expected to invest at least as much in innovation as a challenger. A patent-race game with a stochastic invention process is presented. The result is that when the leading firm deliberately decides to forgo being first-mover in the new market, developing and then ‘shelving’ its new product until the current bestseller is challenged successfully by the rival, the leading firm may spend more than its challengers on R&D, thereby retaining a competitive advantage in innovation of new-generation products. The paper concludes with a discussion of the practical implications of the model.