There has been massive investment in the development of clinical terminologies for use in electronic patient records. However, there has been little published evidence for the added value for primary care that implementation of such a terminology would offer. This paper outlines a methodology that has been used to compare two existing coding schemes (Read codes 5 byte set and Clinical Terms Version 3-CTV3) and demonstrates their relative performance using a certainty-agreement diagram. In the study described, CTV3 offers improved accuracy and consistency with improved usability. The potential advantages of the recently released terminology, SNOMED Clinical Terms, are briefly considered in this context.