Violent discipline in schools infringes on children's rights and is associated with harmful developmental consequences for students. This calls for effective intervention programs, particularly in countries with high prevalence of violent discipline in schools. This study tested the effectiveness of the preventative intervention "Interaction Competencies with Children--for Teachers" ("ICC-T") in reducing violent discipline by teachers in a two-arm matched cluster-randomized controlled trial. The sample comprised teachers (n = 173, 53.7% female) and students (n = 914, 50.5% girls) from 12 public primary schools from six regions in Tanzania. Teacher physical and emotional violent discipline reported by teachers and students was assessed before and six to eight and a half months after the intervention. The schools were randomly allocated to either intervention (6; "ICC-T") or control condition (6; no intervention). Teachers were not blinded. Students and research assistants conducting the follow-up assessment were blinded. A series of multivariate multilevel models revealed significant "time*intervention" effects on physical violent discipline reported by teachers and students and teachers' favorable attitudes towards physical violent discipline, "FDRs" < 0.05. In addition, we found a spill-over effect on peer-to-peer violence and students' externalizing, "ps" < 0.05. There were no significant "time*intervention" effects either on emotional violent discipline, "FDRs" > 0.05, teachers' favorable attitudes towards emotional violent discipline or on student's internalizing problems and academic performance, "ps" > 0.05. Our results provide further evidence that "ICC-T" may positively change teachers' violent disciplining behavior and their attitudes towards violent discipline. ClinicalTrials.gov, NCT03893851.