By developing a simple game in which higher courts and lower courts interact while seeking their own policy goals, this paper examines the reversal behavior of the U.S. Courts of Appeals using the insights of agency theory. Drawing on the payoffs of the circuit courts and the district courts, the author hypothesizes that appellate courts reverse trial courts strategically. When the preference distance between the two courts is small, higher courts are more likely to reverse lower courts decisions that are contrary to lower courts own preferences, but to affirm those similar to the lower courts own preferences. When the distance is large, the higher courts act opposite. The consistency between policy outcomes of lower courts decisions and their preferences, and higher courts reversal propensity are also important factors influencing higher courts behavior. An empirical analysis of Courts of Appeals economic cases over seven years (1993-1999) partially supports these expectations.