This study examines the effects of mixing offenders with non-offenders at Skejby half-way house in Denmark. Skejby half-way house constitutes an outstanding social experiment, as offenders are deliberately mixed with non-offenders in order to reduce their risk of recidivism. The treatment group consists of offenders who have stayed at Skejby half-way house while the controls are selected from residents of four other half-way houses. A Cox regression analysis reveals that the probability of reoffending is 21 percent lower for the treatment group than it is for the control group. (Published Abstract)