Despite some variations among cyberbullying studies, about 20% of the youth population surveyed appears to be involved in cyberbullying. Coincidentally, the current study found that exactly 20% of the students surveyed were involved in cyberbullying as bullies (7.8%), victims (7.5%), and bully/victims (4.7%). What divides those 20% from the 80% of noninvolved students? This study aimed to produce a parsimonious and accurate model that can predict the occurrence of involvement in cyberbullying among youth. Data were collected from a questionnaire survey administered to 1,036 students enrolled in secondary schools in South Korea. Stepwise logistic regression (SLR) was carried out to predict the dichotomous dependent variable (involved/noninvolved) with 10 independent variables grouped into three categories: (a) demographic, (b) media-related, and (c) school and psychology factors. The result of the SLR analysis yielded a four-step model including the variables of cyber-confidence, weekday game time, mobile activities, and age as being significant in explaining the 20/80 division (model χ(2)=34.306, df=4, p<0.001, Nagelkerke R(2)=0.071). This finding suggests that younger students who spend more time playing games on weekdays while being more confident in cyberspace and active in using mobile phones are more likely to be involved in cyberbullying than other students. In particular, the construct of cyber-confidence calls for further elaboration and research, given its controversial function with respect to adolescents' involvement in cyberbullying. Also, this study may bring about insights into practical considerations needed for concerned researchers, teachers, and parents to identify who is inside the group involved in cyberbullying so as to help the participating adolescents escape from the circle of cyberbullying.