A random sample of over 1,000 auto negligence cases filed in 8 courts was examined, including those cases assigned to arbitration and those that were not. In addition, a survey was conducted of about 300 litigents and 400 attorneys involved in arbitration. Results indicate that the program captures a significant fraction of automobile negligence cases, but relatively few are disposed of by an arbitration judgement. About 68 percent were assigned to the program, but many were settled before reaching arbitration. About 55 percent were arbitrated, and about 40 percent were terminated by an arbitration judgement. More than half of arbitrated cases were appealed, but trials de novo were rare. It was not possible to detect a significant effect of the program on trial rates, but rates were initially quite low (less than 5 percent). Cases assigned to the program were more likely to be adjudicated and terminated at a slower rate. While the program slightly increased case activity by attorneys and court staff, it has not had a measurable effect on attorney hours and fees. Finally, disputants and attorneys were favorable in their evaluation of arbitration, with plaintiffs and their counsel being more favorable than defendants and their counsel. Tables, figures, data abstraction forms, and 16 references.