The authors investigated the implementation of the Good Behavior Game (GBG) on disruptive behaviors of high school special education students. Additionally, the study compared the effect of student-selected rewards versus teacher-selected rewards. A multielement withdrawal (A/[B + C]/A/[B + C]) design was used in the classroom to monitor behavior and compare the reward topography. Results suggested that the GBG decreased the frequency of disruptive behavior in both teams; however, there were no notable differences across reward type for either team. Effect sizes were reported using nonoverlap of all pairs. Nonoverlap of all pairs scores indicated moderate-to-large effects for the GBG, but small effects for reward topographies. Limitations and future directions are also discussed.