The bargaining theory of war is a revolutionary breakthrough in theoretical explanations for war in the field of International Relations (IR). This bargaining approach applies game-theoretical deduction to arrive at one proposition that this article focuses on: wars as failed bargains between two rational unitary state actors arise from incomplete information. This provides an important micro-foundation for the use of confidence-building measures to promote peace in foreign policy practices. Yet, the related empirical literature has seldomly tested the direct causal link between information and war in experimental settings. To fill this gap, this study has designed and conducted the first randomized experiment in Taiwan to test the empirical implications of the theory. Through experimentally controlling the personal traits of actors and the distribution of power between them by random assignment, this study finds that war has occurred more frequently and has been more likely to occur between actors as more incomplete information about the distribution of power is revealed to actors in the process of interstate bargaining. These findings are in line with the bargaining theory of war; they also provide more internally valid empirical evidence in support of policy prescriptions for confidence-building measures.