We are in full agreement with Dr Murray regarding the utility of large-scale, population-based studies. These are highly desirable and will, we hope, be facilitated by the recent establishment of the Mental Health Research Network (http://www.mhrn.info) under the auspices of the UK Clinical Research Collaboration (http://www.ukcrc.org). We also agree that longitudinal variables such as course, outcome and treatment response might be key to classification, as Kraepelin supposed. However, although we have not undertaken relevant population studies ourselves, we are not convinced that Kraepelinian dichotomous categories are any more useful in population-based samples than in clinical samples. We find the studies of Van Os and colleagues (e.g. Reference Krabbendam, Myin-Germeys and De GraafKrabbendam et al, 2004) persuasive that dimensional measures are useful in describing psychosis-related morbidity in the general population and, contrary to the proposition of Dr Murray, we would expect dimensions to be more useful than categories in populations unselected for severe illness.

Finally, we would like to restate and further emphasise our optimism about the likely rate of progress in identifying biological markers that can validate psychiatric diagnoses. Markers (in the form of genetic polymorphisms) have already been identified that challenge current nosology. For example, using the Bipolar Affective Disorder Dimension Scale (which rates affective and psychotic dimensions; Reference Craddock, Jones and KirovCraddock et al, 2004) in a study of over 600 cases each of schizophrenia and bipolar disorder, we have demonstrated that a risk variant within the Neuregulin 1 gene, which has been associated with risk of schizophrenia in several samples (reviewed in Reference Craddock, O'Donovan and OwenCraddock et al, 2005), may confer specific risk for a form of psychotic illness characterised by features of both mania and mood-incongruent psychosis (Reference Green, Raybould and MacgregorGreen et al, 2005). Other findings of a similar nature are currently emerging from our own studies and those of other groups, and we anticipate that we are entering a period during which psychiatric research and practice will be placed on much firmer nosological foundations than has been possible in the past.