Abstract:This article reports findings from an empirical study of mediation in domestic relations cases. Cases were screened for domestic violence before mediation, and the mediators could decide to remove cases from mediation if there were concerns about intimidation or safety. Among cases that proceeded to mediation, the likelihood of settlement was not affected by whether the case involved violence, nor by the frequency, severity, or recency of the reported violence. Parties who reported violence either had more favorable assessments of mediation, or their assessments did not differ from those who did not report violence. How the mediator handled the case upon learning of domestic violence affected the likelihood of settlement, such that cases in which the parties were separated for the entire session were less likely to settle.