The aim for this study is to investigate the role of large-scale data in the formation of Swedish upper secondary programme structures, in relation to gender. In the analysis we draw from theories of hegemonic power relations and the concept of assemblage. The results suggest that large-scale evaluations have taken part in forming a two-by two recruitment matrix, where the science and technology areas are valued over "other" areas. By our policy analysis together with our analysis of Swedish registry data, we suggest that the continuous re-organisation of Swedish upper secondary programmes as either "science or social" contributes to the persistent gender biases in student recruitment, and that this should be accounted for in future reforms.