We study countries choosing armament levels and then whether or not to go to war. We show that if the costs of war are not overly high or low, then all equilibria must involve "dove," "hawk," and "deterrent" strategies and the probability of war is positive (but less than one) in any given...