A common argument is that international trade prevents conflict because the possible loss of trade reduces the willingness of both sides to fight. I examine the logic of this argument in the light of game-theoretic models of conflict. In such models, crises are contests of relative resolve. Neither side, however, can observe the other side's resolve in totality; resolve has observable and unobservable components. Instead both sides try to judge the other's unobservable resolve from their actions in the crisis. Costly signals play a critical role in the communication of unobservable resolve in these models. If higher trade flows reduce both sides' resolve for war, then the effect of trade on the likelihood of conflict is indeterminate. Trade flows are observable ex ante, and a state contemplating conflict considers its effect on both sides' actions before beginning a dispute. The initiator is less willing to fight, reducing the chance that it initiates a dispute. At the same time, the target is also less willing to fight, increasing the chance that it makes concessions to the initiator to avoid war, and thus increasing the chance that the initiator begins a dispute. The net effect of these two changes is indeterminate. Trade flows could reduce the risk of escalation by increasing the range of costly signals of resolve in a crisis. A greater range of available costly signals increases the efficiency of signaling between the disputants, increasing the chance that they will reach a peaceful settlement.