Recent research has shown that judges on panels decide cases differently than they do individually. Understanding these panel effects is essential to understanding and predicting judicial behavior. This Article uses a unique naturalexperiment, and interviewsof United States district court judges who participatedin this ex-periment, to empirically investigate panel effects. Specifically, in fourteen district courts the judges chose to sit in an en banc panelto decide the constitutionalityof the FederalSentencing Guide- lines; in fifty-three other districts, the judges decided the issue in- dividually instead. This Article compares the decisions and the characteristicsof these districts to study how panels affect judicialdecision making and to answer the related question of why federal

judges who have the authority to decide a case individually would choose to do so as part of a panel instead.

Among the panel effects the Article finds is that judges in dis- tricts that sat en banc were much more likely to be unanimous intheir voting and also were more likely to find the Guidelines un-constitutional than were judges in other districts. In addition, it appears that a primary purpose of sitting en banc was to obtain these panel effects. Finally, the Article provides evidence of the ef-fects of court structure and composition on judicial collegiality and the propensity to sit en banc. Among the issues the Article ex-amines are how the number of judges and the geographic dis-tances betweenjudges on a court affect judicialcollegiality and thelikelihood that a court sits en banc.