The committee, which was created in 1979, found that much has been written about what makes a good judge but that few specific guidelines, standards, or methods for establishing and measuring valid criteria for judicial performance have been developed. In a broad sense, the evaluation of judicial performance should encompass the selection, evaluation, and discipline of judges. However, little has been done to link the three concepts, either in practice or in the literature. Interest in performance evaluation is likely to increase as States acquire more experience with judicial merit selection systems. Currently used methods of evaluation, such as bar polls, court observation, and news media statements, are usually one-time events which are limited to specific judges. Three jurisdictions, Alaska, the District of Columbia, and New Jersey, have adopted programs requiring regular, formal evaluation of all judges. The New Jersey plan is primarily an internal program consistent with the appointment rather than the election of judges in that State. In the District of Columbia, evaluations are designed to assist the President in making reappointments. Only Alaska goes directly to the voters with evaluation results and specific recommendations about a judge's retention. These approaches formed the basis of the committee's discussions and recommendations. The committee assumes that the public is entitled to more information than it now receives about the judiciary. In addition, the committee believes that the judges themselves should be directly involved in the creation of any evaluation system. The primary purpose of the judicial evaluation should be to improve the judicial system. With these assumptions in mind, the committee surveyed judges for the purpose of identifying criteria that are the most important for evaluating judicial performance. These criteria include quality of reasoning, conscientiousness, legal knowledge, neutrality, commonsense, and intellectual honesty. The committee concludes that judicial evaluation should be a local function; judges should be evaluated by those they most frequently serve. The article provides 11 notes.