Sweden uses municipally run pupil referral units (PRUs) for students displaying emotional behavioural difficulties (EBD). This study investigates one Swedish municipality where transfers of students to PRUs were related to school practices favouring either inclusion or exclusion. A purposeful sampling procedure was used to select three elementary schools with regard to their success (school A) or failure (B and C) in keeping pupils with problematic behaviours at school. These three schools are in catchment areas with similar socio-economic conditions. A mixed methods research design was combined with comparative case studies in a retroductive approach. School A had only one pupil transferred to a PRU in 10 years, whereas such transfers from B and C were almost tenfold. School A distinguished itself from the other two schools by its number of inclusive qualities. This school succeeded in keeping almost all students without depriving other students of their rightful learning.