PURPOSE: This article aims to (1) evaluate whether bullying typology predicts violent injury; (2) longitudinally examine whether violent injury trajectories differ across bullying typology as children age; and (3) longitudinally determine whether children who consistently reported perpetration or victimization (i.e., reported bullying at fifth, seventh, and 10th grade) were different from children who inconsistently reported perpetration or victimization. METHODS: Longitudinal data were obtained from 4,297 children at three waves (fifth, seventh, and 10th grade) in three United States communities. Children were categorized into four mutually exclusive bullying typologies: neither victim nor perpetrator; victim only; perpetrator only; victim-perpetrator. Children self-reported mechanisms of violent injuries that needed medical attention in the past year. Regression models were used to evaluate the relationship between bullying group and the likelihood of violent injury over time. RESULTS: Seventeen percent (n=857) of children in fifth grade reported a violent injury. Prevalence of overall violent injuries, and specifically firearm and knife injuries, increased over time. Children who reported perpetration in the absence of victimization were at increased odds for violent injury (adjusted odds ratio = 1.41, 95% confidence interval: 1.24, 1.60) compared with children who reported neither victimization nor perpetration, while children who reported victimization in the absence of perpetration were at decreased odds (adjusted odds ratio=.84, 95% confidence interval: .73, .97). A significant linear dose-response relationship was observed between duration of bullying perpetration and violent injury. CONCLUSIONS: The relationship between bullying perpetration and violent injury over time was strong. Future research should investigate potential mediating behaviors, such as weapon access, which might explain the observed relationship.