This dissertation investigates the role of power and theory in alternative dispute resolution (ADR), with a focus on mediation. As a scholarly field, mediation presents a heterogenous picture in which notions of expertise, neutral instruments, efficiency arguments and game theory are meshed with psychologically- and cognitively-informed methods that aim to address conflict resolution in a more holistic manner. Although it is a deeply public and political activity, much of mediation theory and practice is framed as normatively neutral, a technical “tool” among many for addressing disputes. More abstract and theoretical debates have largely been confined to critiques of mediation, with the exception of scholarship that uses deliberative democracy. Mediation, especially in its Law School iteration, is a prime example of the lasting influence of legal realism, philosophical pragmatism, and liberal political thought more generally. This has left little disciplinary space for developing a critical and self-reflexive theory of mediation, the politics of ADR, the standard of justice at work in mediation, and the question of power and authority more generally. The dissertation scrutinizes transformative mediation, an outlier to the relative political neutrality, by probing its foundational literature and the translation of its relational worldview in the context of mediator training. It concludes that despite its criticism of liberal norms around individualism and (forced) consensus, TM relies heavily on individual choices and general process belief. Mediation theory rarely addresses the question of power, understood as structural and productive, not only as coercive and institutional. This absence reflects liberal political norms around rationality, proceduralism and control. Turning to critiques of liberal political thought and deliberative democracy drawn from political theory, I argue that mediation is an instantiation of liberalism’s inability to address productive and structural power, and it risks obscuring forms of domination and control by integrating disparate dynamics into one privatized tool. Finally, I point to a different political imaginary that would require theory to be worldly but not pragmatist and that would take the “underlying needs” of a conflict seriously without imbuing them with the pathos of the private.