The purpose of the present study was to develop and validate a scale for measuring children's reactions to peer bullying. The children's reactions scale to bullying(CRSB) consisted of 16 pictures and episodes, each of which described direct or relational bullying situations. Children were asked what they would do if they witnessed peer bullying situations and forced to choose one of four response options for each episode. Four subscale scores were computed by summing the number of each response, retaliating on perpetrators, advocating bullies, watching bullying and doing noting(bystanding), or helping victims. The CRSB and a series of questionnaires were administered to a sample of 266 grade 6 elementary school students to establish its psychometric properties. The results suggested that 4 subscales had good internal consistencies(Cronbach's alpha = .71～.86). Correlation and factor analysis were used to investigate construct validity. There were significant correlations between advocating behavior and bullying experience(r=.36, p<.01), bystanding behavior and prosocial behavior(r=-.30, p<.01), and helping behavior and prosocial behavior(r=.34, p<.01), and bullying experience(r=-.27, p<.01). A three-factor solution emerged, consisting of a bystanding/helping, retaliating, and advocating behavior factor. These factors accounted for 36.5%(bystanding), 24.30%(retaliating), and 21.07%(advocating) of the variance on the CRSB, respectively. When children were categorized as ‘bullies’, ‘bully/victims’, ‘pure victims’ or ‘not involved’ on the basis of Social Experience Questionnaire, bullies were less likely to help the victims and more likely to advocate the perpetrators than other children. Limitations of the current study and implications for future research were discussed.