In the UK, as in several developed countries, concern has been expressed by interested commentators about the apparent decline of post-16 participation in the "hard" sciences (especially physics and chemistry). While formal full-time participation in 16-19 education and higher education has increased since the 1990s, both the relative and absolute numbers studying physics and chemistry have declined. If the development of scientists is seen as key to economic, technical and intellectual progress then this decline could be very serious indeed. One way of understanding and perhaps remedying this decline is to consider those currently under-represented. In recent years much attention has focused on differences in participation by boys and girls. However, this paper examines the changes in post-compulsory science participation since the 1990s in terms of socio-economic factors to answer the question--is science a middle-class phenomenon? It uses Pupil-level annual schools census/National pupil database (PLASC/NPD) and Higher Education Statistics Agency/Universities and Colleges Admissions Service (HESA/UCAS) data sets, and summarises a review of 1083 pieces of relevant literature, conducted by the authors for the Royal Society. (Contains 5 tables.)