A distinctive characteristic of our mechanisms for conflict resolution is that litigation is carried out by agents chosen by disputing principals. Does the fact that clients choose lawyers to carry on their disputes facilitate dispute resolution or instead exacerbate conflict? The dominant contemporary view is that the involvement of lawyers magnifies the contentiousness of litigation and wastes social resources, prolonging and escalating the conflict in ways that enrich the legal profession but not the clients. But in a recent article, Ronald Gilson and Robert Mnookin suggest another possibility: by choosing lawyers with reputations for cooperation, clients may commit to cooperative litigation in circumstances where the clients themselves would not otherwise trust each other. Using the methodology of experimental economics, this article presents a test of their idea that, by choosing cooperative agents under well-specified procedures, principals may sustain more cooperation than they could on their own. Our experimental findings are consistent with the Gilson-Mnookin hypothesis.