This study assesses several policy implications of within-school, between-classroom variability in pupil achievement. It diverges from current school effect studies by directly modelling pupil achievement in the Jerusalem public primary school system. This three-level study includes pupils, classrooms and schools, thus allowing an appropriate estimate of the variations between these three levels. The findings show that between-classroom variability is consistently greater than the estimated variation between schools. These findings contrast with traditional school-level analyses that usually ignore within-school variability. In the light of these findings we address three educational and policy issues. First, we probe into the moral consequences of between-classroom, within-school variability, specifically focusing on issues of choice and commitment. Second, we scrutinize the administrative policy of ‘social integration’ and reflect on some educational consequences that result from our findings. Third, we assess the Israeli version of ‘school league tables’ and discuss their usefulness as a means of resource allocation.