After over 20 years of fighting in Northern Ireland, the Belfast Agreement of 1998 has successfully implemented a power-sharing agreement. Belfast was not the first attempt at a peaceful settlement; indeed, some scholars count as many as seven prior peace attempts in Northern Ireland, the most significant being the Sunningdale Agreement of 1973. Why was it that politicians successfully negotiated the peace in 1998, while these prior attempts failed? Drawing from social network theory, I argue that the Belfast Agreement succeeded, not because of a change of interests or disappearance of spoilers, but because of the presence of brokers at the bargaining table. Brokers, in particular, have the capacity to legitimate settlements—to frame settlements in such a way that they appear consistent with principles held by multiple coalitions. As a result, brokers are both more likely to build a winning coalition for a settlement, as well as marginalize spoilers who seek to undermine the peace.