Three important sets of empirical questions are isolated: whether the jury is a competent fact finder, able to assess evidence, understand legal instruction, and reach appropriate results; whether juries are representative of the communities from which they come and, where not, whether juries or representative composition would achieve significantly different verdicts; and whether non-unanimous juries and juries smaller than twelve are inferior to traditional twelve-member juries. Findings on all these questions are inconsistent and in dispute. All of the empirical research is flawed by severe methodological limitations, but faulty interpretation of research results is a more important problem. A critical evaluation of the empirical research casts doubt on many of the claims made by researchers about the working of the jury. Part of the confusion results from different notions of the jury's role. Of the five commonly employed research methods, this article focuses on the adoption of various forms of simulated jury panels designed to reconstruct or parallel the deliberations of the real jury and on attempts to assess jury performance through the views of other participants in the trial. These are the methods upon which most research efforts, both in Great Britain and the U.S., have been expended and upon which an evaluation of jury performance must depend. In one study as many as a quarter of the verdicts of the shadow juries were different from those of the real jury, however. In addition, student juries do not even approximate the social mix which would characterize any real jury. Examination of over 300 jury verdicts in Birmingham, England, produced results comparable to those of other U.S. studies. Findings showed that individual attitudes and prejudices probably exerted greater influence on the final verdict than did any general social characteristic of the jury as a whole. In both Great Britain and the U.S., the introduction of majority verdicts, administrative meddling with jury venire, increasing attempts to handpick jurors, and, above all, the acceptance in the U.S. of juries of fewer than twelve members, drastically alter traditional notions about jury trial. Future research should consider, among other subjects, whether juries are capable of accurately discriminating between the factually innocent and the legally and factually guilty; whether juries consciously and routinely take into account equitable factors; and whether defendants prefer open-court adjudication to other methods of dispute resolution such as bench trial or plea bargaining. Over 50 footnotes and 175 references are appended. (Author abstract modified).