For the past 30 yr wave—swept shores have served as a model system for experimentation in community ecology. Due in large part to the severity of the physical environment, individual plants and animals are frequently disturbed, turnover in the community is rapid, and experiments can be conducted in months which in other habitats would require years. However, the experimental advantage of rapid turnover must be weighed against our ability to account for its causes. Only if we can predict the rate of turnover can we predict the dynamics of the community. On wave—swept shores where disturbance is dominated by environmental effects, the ultimate ability to predict community structure rests on the the proximal ability to predict the physical environment and to understand its consequences. Some important aspects of the wave—swept environment (such as the tides) are well understood, but the effects of wave—induced hydrodynamic forces, perhaps the predominant environmental stress on shoreline organisms, has been thought to be unpredictable. Indeed, the stochastic nature of ocean waves precludes the short—term prediction of wave forces. However, as with the random motion of molecules in a gas, the short—term unpredictability of the ocean's surface can form the basis for a robust statistical approach to the prediction of long—term events. This study employs the statistics of the random sea to predict the largest wave to which a littoral site will be subjected in a year (≈5.9 times the yearly average significant wave height), and uses hydrodynamic theory to predict the force that this large wave exerts on individual organisms. The result is a quantitative measure of wave exposure, a mechanistic link between local wave climate and species—specific survivorship that can be used as a tool for exploring the relationship between environmental severity and community ecology. The proposed method is tested by predicting the rate at which patches of bare substratum are formed in beds of the mussel Mytilus californianus, a dominant competitor for space on rocky shores in the Pacific Northwest. Predicted rates are very similar to those measured in the field, suggesting that this method can provide useful input into models of intertidal patch dynamics. Data from several sites around the world suggest that the yearly average waviness of the ocean at any particular site can (over the course of decades) vary by as much as 80% of the long—term mean. The methodology proposed here allows this decade—to—decade variation in wave climate to be translated into the resulting variation in survivorship; predicting, for example, that an increase of 1 m in yearly average significant wave height results in a fourfold increase in the rate of patch formation in a mussel bed. Such a shift would have substantial consequences for community dynamics. M. californianus is unusual in that the expected wave—induced stress is near the species' modal strength, and it will be of interest to determine if this characteristic is common among dominant competitors for space. To ease the application of this method by ecologists, a list of hydrodynamic force coefficients is provided to allow the method to be used with a wide variety of plants and animals, and several sources of information regarding wave heights are suggested. The assumptions, limitations, and procedures for practical application of the approach are discussed. The method described here for wave—swept shores can, in theory, be applied to the mechanistic study of physical stresses in other environments. In many habitats, however, multiple stresses (rather than one dominant stress) have strong effects on community dynamics, rendering prediction difficult. In this regard, the practical advantage of the wave—swept shore, where water velocity is of such overarching importance, is again evident.