Although the diversionary theory of international conflict is intuitively plausible and is supported by much anecdotal evidence, quantitative tests generally have failed to establish a systemic link between domestic political troubles and foreign policy adventurism. In this article we revise the diversionary theory. We argue that a state leader will treat an erosion of domestic support more seriously when it comes from within segments of society that are critical in the maintenance of the leader's ruling coalition than when it comes from other domestic groups. We also argue that diversionary behavior typically should involve actions short of war. The revised theory is tested for the United States over the period from 1953 to 1976. Multiple empirical techniques are used to analyze the relationship between the uses of external force by the United States and the level of support for the president among members of the president's party. The results provide strong support for the revised diversionary theory.