While individual data form the base for much empirical analysis in education, this is not the case for analysis of technical efficiency. In this paper, efficiency is estimated using individual data which is then aggregated to larger groups of students. Using an individual approach to technical efficiency makes it possible to carry out studies on a wide range of student constellations; e.g. entire educational programmes, schools or smaller groups of students within a school or a programme. Efficiency is estimated for students graduating from Swedish upper secondary school in 1999, and aggregated to six programmes and to male and female students separately within each programme. Efficiency differences are related to gender distribution within the programmes and to the students' effort allocation between subjects. We find no general trend among the programmes that gender distribution should be related to efficiency, but the effort allocation is related both to gender and to gender distribution within programmes.