Identifying diseases by means of their characteristic symptoms and signs is one of the cornerstones of medicine whose value is usually taken for granted. It therefore comes as something of a surprise to find that the concepts of disease and diagnosis are fraught with difficulties and uncertainties (e.g. see Kendell, 1975; Clare, 1980). Every attempt to capture the essential nature of disease – as suffering, as presence of a lesion, as biological disadvantage, and so on – has been found to be unsatisfactory; like health, disease eludes easy definition. Diagnosis is also a rather less straightforward process than it might appear. The theoretical classification of disease, just like the classification of any natural phenomena, can never hope to be perfect, and there will always be areas of dispute. The practical act of making a diagnosis is less perfect still, being subject to human error and, rather more than might be wished, to the vagaries of medical fashion.