The idea of a comprehensive coding system, able to code any concept has allure; similarly linking a term to related concepts. However, there appear to be flaws: firstly just getting these links right and keeping them up-to-date, secondly how you find data in such a comprehensive system? If limited lists have to be imposed to make things usable why not just stick with the limited list? Routine data offers enormous opportunities for quality improvement and research;12 we should avoid putting this at risk until these issues are resolved. Perhaps a head-to-head test of SNOMED CT v. the widely used combination of the International Classification of Disease (ICD)13 and the International Classification of Primary Care (ICPC)14,15 is needed to resolve these important issues. Further debate is needed about the pros and cons of SNOMED CT. Are these teething problems or fundamental flaws?