Studies of the trade-conflict relationship typically emphasize dyadic over multilateral trade, ignoring the large-scale effects of trade integration. Openness, a common measure of integration, is conceptually problematic and yields inconsistent empirical results. Drawing on the concept of network centrality, I develop a network approach to integration, emphasizing three aspects of multilateral trade: breadth of trade ties, depth of trade ties, and commercial distance between nonpartners. I then extend the dyadic logics of the trade-conflict literature to the multilateral level, focusing on (1) theories that predict that trade reduces conflict, either by increasing opportunity costs or by creating signaling mechanisms, and (2) theories that predict that trade, especially when asymmetric, increases the political autonomy of states and thus encourages aggression. Extensive country-year analysis shows that, consistent with the first set of theories, network centrality unilaterally constrains aggression. More central states initiate fewer conflicts, even when their trade ties are highly asymmetric.