The quality of primary care should ultimately be judged on the effect of health outcome of individual patients. However, for the foreseeable future, it is inconceivable that the necessary data will be available to implement this principle. And in any case, specification of the necessary statistical model is fraught with difficulty. This paper therefore applies data envelopment analysis (DEA) to quality in primary health care administration, in the belief that it offers a consistent and helpful “intermediate technology” for assessing performance. Many of the outputs of primary care are intangible and unfold over a long time period. It is therefore useful to judge the quality of primary care on the basis of process variables, which are particularly well suited to DEA. Moreover, DEA is not vulnerable to the misspecification bias that afflicts statistical models. The principle difficulty DEA gives rise to is the selection of relevant environmental variables. The issues are illustrated with an example from English Family Health Service Authorities.