Bargaining models play a central role in international relations, particularly in the study of conflict. A common criticism of this approach is that it fails to account for nonmaterial (e.g., psychological) factors that may influence the bargaining process. We augment a standard bargaining model by allowing actors’ preferences over conflict to diverge from the “fitness” payoffs (e.g., resources) typical of such models. Preferences are subject to evolutionary forces—those who attain high fitness reproduce more. We find that (1) there is a trade-off where being “irrationally” tough leads to better bargains but also more inefficient conflict; (2) actors develop behavioral biases consistent with empirical findings from psychology and behavioral economics; and (3) these behavioral biases inevitably lead to conflict. By bridging the strategic and psychological approaches to conflict, our models provide new insights into questions such as how changes in military and intelligence-gathering technology affect the likelihood and expected cost of war, and how to interpret the purported decline of violence over recent human history.