OBJECTIVE--To compare the clinical efficacy, patient satisfaction, and cost of three specialist treatments for depressive illness with routine care by general practitioners in primary care. DESIGN--Prospective, randomised allocation to amitriptyline prescribed by a psychiatrist, cognitive behaviour therapy from a clinical psychologist, counselling and case work by a social worker, or routine care by a general practitioner. SUBJECTS AND SETTING--121 patients aged between 18 and 65 years suffering depressive illness (without psychotic features) meeting the criteria of the Diagnostic and Statistical Manual of Mental Disorders, Third Edition for major depressive episode in 14 primary care practices in southern Edinburgh. MAIN OUTCOME MEASURES--Standard observer rating of depression at outset and after four and 16 weeks. Numbers of patients recovered at four and 16 weeks. Total length and cost of therapist contact. Structured evaluation of treatment by patients at 16 weeks. RESULTS--Marked improvement in depressive symptoms occurred in all treatment groups over 16 weeks. Any clinical advantages of specialist treatments over routine general practitioner care were small, but specialist treatment involved at least four times as much therapist contact and cost at least twice as much as routine general practitioner care. Psychological treatments, especially social work counselling, were most positively evaluated by patients. CONCLUSIONS--The additional costs associated with specialist treatments of new episodes of mild to moderate depressive illness presenting in primary care were not commensurate with their clinical superiority over routine general practitioner care. A proper cost-benefit analysis requires information about the ability of specialist treatment to prevent future episodes of depression.