The learning curve has become a central concept for corporate strategic planning. However, strategies based on the learning curve often fail to achieve their intended results. This paper explores the implications of the learning curve for competitive strategy under a range of assumptions regarding competition and the nature of the learning process. The paper first considers the optimal decision rules which apply when a learning curve is present. A dynamic model of industry equilibrium is then used to study how the rate of learning and information diffusion affect entry barriers, profits, and the time path of price and output. The results show that entry barriers are quite high when learning remains proprietary but are substantially reduced when learning diffuses across firms. With proprietary learning, the firms optimal pricing policy is to set initial price below current cost and hold price approximately constant as cost falls over time. Diffusion of learning shifts the optimal policy back toward the classical MR = MC rule, based on current marginal cost. This causes price and cost to fall roughly in parallel, as observed in most empirical studies of long-run price behavior. The incentives which arise from the learning curve tend to intensify competition and reduce industry profits, except in rare cases where preemption is feasible. Profits are higher if learning diffuses across firms, or if all firms act myopically. In general, the results highlight the importance of information diffusion and the danger of following simple strategy prescriptions based on the learning curve.