We investigated the effects of classroom separation on twins' behavior, progress at school, and reading abilities. This investigation was part of a longitudinal study of a nationally-representative sample of twins (the E-risk Study) who were assessed at the start of school (age 5) and followed up (age 7). We examined three groups of twins: pairs who were in the same class at both ages; pairs who were in separate classes at both ages; and pairs who were in the same class at age 5, but separated by age 7. When compared to those not separated, those separated early had significantly more teacher-rated internalizing problems and those separated later showed more internalizing problems and lower reading scores. Monozygotic (DZ) twins showed more problems as a result of separation than dizygotic (DZ) twins. No group differences emerged for externalizing problems, ADHD or prosocial behaviors. The implications of the findings for parents and teachers of twins, and for school practices about separating twins, are discussed.