Fifty-three cases are reported. Thirty-five were offered psychotherapy but 10 left treatment early. Nine patients refused psychotherapy and nine were unsuitable. Outcome was measured by change in contacts with local health services and general practitioners for an equivalent time period before and after psychotherapy or assessment. Of the treatment group, 17 recovered, three were improved, one was unchanged and four were worse (P << 0.00 1). There was a reduction in general practice use by those who received even brief treatment. Those with anxiety symptoms improved most. The ‘refuser’ group contained more depressed patients and patients with alcohol or personality problems did less well. Variables not associated with outcome were: duration of treatment, previous treatment, therapist experience, therapist gender, number of cases previously seen and premature termination. Therapist prediction of outcome was not usually valid. Costs compared favourably with the costs of other forms of treatment.