This study has two distinct features that are different from current studies using multilevel analysis on school effectiveness. First, it uses a very large dataset, and second, the analysis is on a streamed-school system. The initial findings of this study record three differences from studies in the West: (a) the relatively large size of school effect in the Hong Kong schools; (b) the negative correlation between intercepts and slopes for prior attainment; and (c) the differential effectiveness for low and high ability students. The study highlights the issue of whether or not the results of school effectiveness research based on mixed ability schools can be generalised to some countries in East Asia where students are streamed to schools according to their academic ability. It suggests the need to explore whether the contextual factors that influence the effectiveness of schools in a mixed system are duplicated in a streamed-school system.