Objective: Longitudinal trajectories of bully perpetration were examined across 6 waves of data among adolescents from middle to high school, predictors of these trajectory membership, and outcomes associated with trajectories at Wave 6. Method: Participants completed self-report surveys (6th through 10th) grade. Bully perpetration was input to the trajectory analyses. Static predictors of trajectory membership included gender, positive and negative family relations, and exposure to an intervention. Dynamic covariates of bully trajectories included empathy, impulsivity, depression, and victimization. High school outcomes included delinquency, affiliation with deviant peers, and school belonging. Results: Group-based semiparametric mixture modeling yielded 5 distinct trajectories of bullying perpetration emerged: (a) Low (37.8% of the sample); (b) Moderate Flat (51.3% of the sample); (c) High Declining (3.4%); (d) Middle School Peak (4.2%); and (e) Moderate Escalating (3.4%). Early family relations and intervention status were found to be predictors of trajectory membership. Peer victimization, impulsivity, and depression as dynamic covariates predicted positive deviations from the bullying perpetration trajectories (i.e., increases in bullying), while empathy predicted negative deviations. Trajectory membership was differentially predictive of antisocial behavior, peer affiliation, and school belonging outcomes in the 10th grade. Findings suggest middle school students exposed to a social emotional learning program were less likely to belong to the more serious trajectories, bullying perpetration levels were generally highest during the middle school period, and some bullying perpetration continued into high school. Conclusion: Interventions for both middle and high school need to focus on individual and contextual factors identified in this study. (PsycINFO Database Record (c) 2017 APA, all rights reserved)