Why do successful organizations often move in new directions and then fail? We propose that this pattern is especially likely among organizations that have survived a history of competition. Such experience adapts organizations to their environment, through so-called Red Queen evolution, but being well-adapted for one context makes moving into new contexts more hazardous. Meanwhile, managers in such organizations infer from their histories of competitive success a biased assessment of their organizations dynamic capabilities. Consequently, although surviving competition makes organizational change especially hazardous, managers in surviving organizations are especially inclined to such initiatives. We develop these ideas in an empirically testable model, and find supportive evidence in estimates of the model using data from the history of the U.S. computer industry.