The author shows that inferences about jury behavior cannot be drawn from verdict statistics alone because very plausible rival explanations of jury behavior cannot be eliminated. Most sources of verdict statistics provide information only about cases that reach trial. Depending on case type, jury trials may account for only between 1 and 12 percent of cases filed in court. The remaining cases are either settled, dropped by the plaintiff, or disposed of through judicial or administrative action. In addition, jury behavior changes over time and varies across jurisdictions. Verdict differences may result from the fact that juries are hearing very different evidence or are responding to different substantive law. Jury behavior is also affected by race and gender and by differences in the way lawyers try cases. The "pain and suffering" component of jury awards is one of the most criticized aspects of jury behavior. Wide variations in jury estimates of these awards may be due to unrelability of the subtrahend figure and not to jury behavior. Although substantial methodological problems are associated with the use of verdict statistics, such statistics can provide some valuable information on jury behavior. 60 references, 12 footnotes, and 2 tables