Cost functions are estimated, using both random effects and stochastic frontier methods, for institutions of higher education in England. The paper advances on the existing literature by employing finer disaggregation by subject, institution type, and location, and by introducing consideration of quality effects. The findings are that, amongst undergraduates, medical students are the most costly, and non-science students the least; amongst postgraduates, those on taught courses are costly, while research students are relatively inexpensive. Provision in London is found to be more costly than that elsewhere. Estimates of economies of scale and economies of scope vary according to the choice of estimating technique. The random effects model suggests that ray economies of scale and economies of scope are ubiquitous. The stochastic frontier model suggests some product-specific economies of scale in research, but diseconomies elsewhere, and product specific economies of scope in undergraduate science, but diseconomies elsewhere. This has implications for achieving any expansion in higher education.