The bulk of public debate on education focuses on schools and school differences. Ideally, the characteristics of schools that add value to student performance can be identified and implemented for other schools. However, such scenarios assume that school effects are sizable, stable across cohorts, and consistent across subject areas. This study tests these assumptions by analysing school effects in both primary and secondary schools in 5 achievement domains with administrative data from almost all government school students in Victoria, Australia. Gross school effects are reasonably large but show only moderate stability. However, in net progress models which control for prior achievement, school effects are substantially smaller, display only low levels of stability across cohorts, and are not consistent across achievement domains. Therefore, it is difficult to identify schools that consistently increase (or decrease) student performance across subject areas beyond that expected by students' intake characteristics, most notably prior student performance. Other policy goals are recommended.