Book excerpt: While observers agree that courts function poorly in developing countries, a lack of data has limited our understanding of the causes of malfunction. We combine data from administrative records on severance cases filed in the Mexico City Labor Court with interventions that provide information to parties in randomly selected cases on predicted case outcomes and conciliation services. We first use the data to document a set of stylized facts about the functioning of the court. The interventions nearly double the overall settlement rate, but only when the plaintiff herself is present to receive the information directly. Administrative records from six months after the treatment indicate that the treatment effects remain unchanged over that period, even though an additional one in three cases in the control group settle in that period. The post-treatment results indicate that lawyers do not convey the information provided in the intervention to their clients. A simple analytic framework rationalizes the experimental results. Analysis of settlements induced by the interventions suggests that the provision of information is welfare-improving for the plaintiffs. The experimental results replicate over two phases conducted in different sub-courts, showing robustness.