Objective: To identify severe bullying risk factors in three highland Peruvian zones private school students. Design: Survey type study. Setting: Faculty of Medicine, Universidad Nacional Mayor de San Marcos, Lima, Peru. Participants: Fifth elementary school to fifth high school private school students. Interventions: A survey validated in previous studies to identify school violence (bullying) was applied to 736 students from Ayacucho, Huancavelica and Cusco (Sicuani) private schools between fifth year of elementary school and fifth year of high school. We considered severe bullying when they answered positively five or more of the nine questions referring to types of bullying. We compared 37 students with severe bullying and 74 students randomly selected from the group not having bullying. With bivariate analyses and multivariate logistic regression analysis we identified factors significantly associated to severe bullying. Main outcome measures: Risk factors for severe bullying. Results: Severe bullying was significantly associated to parents’ reaction when knowing the fact, repeated threatening in spite of bullying communication, presence of school gangs, having gang friends, and possessing a physical defect. Logistic regression found significant association with parents’ reaction, gang presence, and possessing a physical defect. Conclusions: Severe bullying is associated to multiple risk factors that can and must be detected early because of the psychological damage generating in the students.Objective: To identify severe bullying risk factors in three highland Peruvian zones private school students. Design: Survey type study. Setting: Faculty of Medicine, Universidad Nacional Mayor de San Marcos, Lima, Peru. Participants: Fifth elementary school to fifth high school private school students. Interventions: A survey validated in previous studies to identify school violence (bullying) was applied to 736 students from Ayacucho, Huancavelica and Cusco (Sicuani) private schools between fifth year of elementary school and fifth year of high school. We considered severe bullying when they answered positively five or more of the nine questions referring to types of bullying. We compared 37 students with severe bullying and 74 students randomly selected from the group not having bullying. With bivariate analyses and multivariate logistic regression analysis we identified factors significantly associated to severe bullying. Main outcome measures: Risk factors for severe bullying. Results: Severe bullying was significantly associated to parents’ reaction when knowing the fact, repeated threatening in spite of bullying communication, presence of school gangs, having gang friends, and possessing a physical defect. Logistic regression found significant association with parents’ reaction, gang presence, and possessing a physical defect. Conclusions: Severe bullying is associated to multiple risk factors that can and must be detected early because of the psychological damage generating in the students.