This paper estimates how peers' achievement gains are affected by the presence of potentially disruptive and emotionally sensitive children in the school-cohort. We exploit that some children move between schools and thus generate variation in peer composition in the receiving school-cohort. We identify three groups of potentially disruptive and emotionally sensitive children from detailed Danish register data: children with divorced parents, children with parents convicted of crime, and children with a psychiatric diagnosis. We find that adding potentially disruptive children lowers the academic achievement of peers by about 1.7-2.3 percent of a standard deviation.