School bullying has become a public health issue in the past several decades which caused physical and psychological harm on bullies, victims as well as bystanders. Therefore, a series of anti-bullying programs were designed to reduce the bullying or the harmfulness of bullying, but the effectiveness of these programs was inconsistent. Forty-three pre-post randomized controlled trail studies were selected to examine the effect of school-based anti-bullying programs. The evidence quality was screened under the GRADE protocol. Results suggested a significant effect for anti-bullying programs which targets bullies (g = 0.57, p< 0.05), while those who target victims, victimization decreased (g = 0.40, p< 0.05), and overall metal health improved (g = 0.40, p< 0.01). The pooled effect size indicated that school-based anti-bullying programs have a small effect on bullying and victimization (g = 0.17, p< 0.001) and influenced by publication bias. Attitude changed moderately (g = 0.19, p< 0.001). The evidence of attitude change was moderate, other outcome variables have a very low or low evidence quality according to GRADE protocol. For the programs targeting universal students, programs which was less than one semester performed better, and the global intervention showed more effective than education. The effect size of anti-bullying programs targeting bullies or victims was moderate, but more research is needed. Though school-based anti-bullying programs which target universal students have a relatively small effect size, however, they are practical for real-world applications. The dose of intervention and the feature of programs have influence on the effect of anti-bullying programs.