At a secondary school in Melbourne a majority of the mathematics teachers favoured the introduction of streaming in Year 9 and 10 Mathematics to replace the established mixed ability system. This study was done in conjunction with the changeover with the intention of providing a measure of the advantage (if any) to be gained from streaming. It covered two topics in the normal mathematics programme and focused on achievement and attitude. Before each topic pretests of achievement were given on which students were ranked into three ability levels, and then randomly assigned to streamed or mixed groups. Teachers were also randomly assigned to the classes and then reassigned for the next topic. Teaching was as near as possible undifferentiated across groups, levels, and years. The findings indicated that no significant gains or losses in achievement could be attributed to streaming across ability levels. Attitude measures were made using the semantic differential technique but comparisons yielded no significant differences or patterns of difference between streamed and mixed groups.