A growing body of work seeks to explain the lack of clear evidence for the diversionary use of force by casting doubt on such strategies' attractiveness for policy makers: while domestic political and economic problems may provide incentives for diversion, such strategies involve political and military risks that frequently outweigh these incentives. Such theories correctly identify the objective risks involved in diversion but do not account for variation in leaders' risk-taking propensities. We develop a “first image” theory of diversion that suggests a key psychological variable (locus of control) shapes leaders' willingness to engage in risky diversionary strategies. A statistical analysis of the American use of force, 1953–2000, finds strong support for this model. We conclude that the lack of clear evidence for diversion in general is a reflection of the contingent nature of the phenomenon and call for greater attention to how agents and structures interact to produce policy behavior.