The treatment of anxiety is one of the more contentious issues in therapeutics. It is also one of the most important, as pathological anxiety is such a common symptom in the population. The recent Office of Population Censuses and Surveys survey of psychiatric morbidity indicated that one in eight of all those interviewed had an anxiety disorder of sufficient severity to receive a formal diagnostic label in the week before interview (Meltzer et al, 1994). Clearly, only a proportion of these should be treated with drugs, and selection for pharmacotherapy is one of the more difficult issues in clinical practice. It is a matter of some concern that changes in the National Health Service are increasingly shifting the onus of responsibility for treatment from the psychiatrist to the general practitioner. As a consequence, it behoves both disciplines to keep liaison active, so that best practice can be maintained from whatever source it is being provided.