National value-added datasets have recently become available that record a pupil's progress from Key Stage 2 right through to GCSE. Such a dataset is clearly a useful tool for assessing the impact various characteristics of secondary schools have on pupil performance. This paper reports on a research project which involved the use of a variety of statistical techniques, including multi-level modelling, to explore the impact of single-sex education and school size on pupils' progress and opportunities. The analysis suggests that single-sex education is particularly beneficial to girls in comprehensive schools and to boys in selective schools. Furthermore, girls' schools were found to counter traditional sex-stereotyping in subject choices. In terms of school size, the results show that medium-sized schools obtain slightly better results than very large or very small schools. The paper concludes by exploring alternative explanations of the identified effects.