The present volume contains 30 chapters by a variety of experts, dealing with conceptual issues that need to be considered in preparation for the next revision of our classifications. The great change in psychiatric classification came with DSM–III, which consisted of 265 mental disorders and replaced clinical descriptions – where the task of the clinician was to recruit the patient to the nearest description – with the Chinese menu, now familiar to us all. Lee Robins pointed out that the rule laid down was that any given symptom could only appear in one disorder. She concluded: ‘I thought then, and I still do, that the rule was not a good one, because it deviates from the practice in the rest of medicine, where many diseases share symptoms’ (p. 268). Thus, although previously anxious symptom had been seen as an integral part of what was then ‘neurotic depression’, anxiety now had to be reassigned to anxiety disorders. Sadly, the arbitrary diagnostic rules built into our classification systems impose tunnel vision on many clinicians, who tend to reify the disorders described and no longer appear to notice symptoms which are there before them. Since then, successive versions of the DSM have added 89 new disorders, and abandoned diagnostic hierarchies, so giving birth to ‘comorbidity’. Recently work has been divided into topic groups, but ‘each working group was reluctant to give up their rights to a particular domain, even when it might have been better categorised elsewhere’ (p. 61).

When the American Psychiatric Association began to consider changes in preparation for DSM–V the problems seemed to be that many patients were found to have multiple comorbidity, that many more were diagnosed as ‘not elsewhere classified’, and that the categorical dichotomies of the DSM system might be supplemented by a dimensional system to allow various degrees of severity of a disorder to be recognised. Ortigo and others argue for a prototype diagnostic system, where each diagnosis would be rated on a 5-point scale, ranging from a poor to a perfect match to a prototype (p. 377).

Maj (p. 263) considers two questions, whether mental disorders are really as common as community surveys suggest that they are, and whether comorbidity can really be so common. He argues that there can be no firm answer to the first problem, and makes cogent objections to the latter. Zachar & Kendler (p. 127) distinguish between ‘disease realists’, who consider that there are qualitative differences between true diseases and normality, between which Nature has beneficently provided joints, and empiricists, who reject these assumptions, and seek to make connections between observable phenomena. For them, validity refers not to whether a disease is really there, but to what kinds of inferences one can make about a patient on the basis of a particular diagnosis. The differences between these fundamentally different approaches to classification echo throughout the volume.

Many of the papers by psychologists clearly take the latter approach, for example those by Krueger's group on the metastructure of the diagnoses produced by the DSM–IV system. Yet unless the metastructure can be radically simplified the comorbidity problem is insoluble, and rival working groups will jealously hold on to their symptoms. The editors do not attempt to draw any general conclusions at the end, and indeed it would be impossible to do so.