Objectives: In order to promote mindfulness in primary school, the Breathing Break Intervention was developed. This collection of short daily breath-based mindfulness practices was introduced to 15 teachers who delivered them up to 3 times a day to their students. Method: In a randomized controlled trial, 146 third and fourth graders (49% female) either received the intervention (n = 81) or participated in the active wait list control group (n = 65). Students were asked to nominate prosocial peers and to report on supportive peer relationships in their classrooms before (pretest) and after (posttest) the 9 weeks of the Breathing Break Intervention, and in a follow-up 5 months later. Results: Mixed multilevel models indicated a group × sex × posttest interaction (t(211) = 2.64, p < 0.01) suggesting that girls in the intervention group were rated to be more prosocial by their peers at posttest than at pretest and than girls in the active control group when children’s age and parents’ education were accounted for. Supportive peer relationships in the active control group deteriorated between pretest and posttest, which occurred immediately before the second school lockdown due to the COVID-19 pandemic, whereas they remained the same in the intervention group (t(223) = 2.56, p < 0.05). Both effects were not maintained at follow-up, probably due to children’s irregular school attendance during the lockdown. Conclusions: Introducing a short daily breathing practice in primary school classrooms seems to be effective in maintaining supportive peer relationships and in stimulating girls’ prosocial behavior. Preregistration: The study was preregistered at aspredicted.org (#44925).