This study investigated whether social position (e.g., gender, migration, family status), intrapersonal-level (e.g., online risk behaviors, motives of Internet use), interpersonal-level (e.g., victimization and bullying), family-level (e.g., parental mediation), and class-level (e.g., teachers' mediation, ethnic diversity) variables predict bias-based cybervictimization. Self-report questionnaires were completed by 1,018 Austrian adolescents (52.3% girls), aged 12 to 17 years ([X-bar] = 13.55, SD = 0.88). The logistic part of a multilevel zero-inflated Poisson model showed that higher levels of offline victimization and a higher proportion of immigrants in classes were predictors for students reporting at least one form of bias-based cybervictimization. The Poisson part of the model showed that being a girl, higher levels of cybervictimization, lower levels of avoiding online risks, and more discussions about media use with teachers in classes were predictors for students reporting a higher number of bias-based cybervictimization. Implications for prevention are discussed.