The existence of significant disparities between the prescribing costs of dispensing and non-dispensing general practices has long been suspected, and received confirmation in a study of Lincolnshire, based on data for 1990-91. This study subsequently attracted much criticism. In this paper, we extend the analysis by considering annual cost and other prescribing data for Lincolnshire for the years between and including 1990-91 and 1993-94, in the light of the prescribing criteria developed by the Audit Commission. Our results show that dispensing practices had higher prescribing costs per patient for all the years analysed. In 1993-94, dispensing practices prescribed more items per patient (fewer of them generically) and were less capable of remaining within their prescribing budgets. The essential difference in prescribing costs lies in the area defined by the Audit Commission as 'core prescribing'. Using the Audit Commission's criteria for 'rational' prescribing, dispensing practices could make a significantly higher level of savings than non-dispensing practices. The findings lend support to the hypothesis advanced in the earlier analysis, namely, that the higher costs of prescribing in dispensing practices are accounted for primarily by management practice and the structure of incentives.