There is compelling evidence from specialist research centres for the efficacy of cognitive-behavioural therapy (CBT) with a variety of psychiatric disorders but effectiveness in routine clinical practice has yet to be established. The complex tasks confronting those generalist practitioners who see a broad range of referrals are analysed in terms of the breadth of treatable disorders, the variable nature of patient characteristics and the pressures of work. This analysis suggests that effectiveness and efficiency in routine practice of CBT will be compromised unless therapists work in small teams and adopt some of the rigours and objectivity of the research context. In particular, it is suggested that clinical practice should include routine assessment of complexity and severity of problems at a screening interview and quality of collaboration after a trial period of treatment. Therapy for those cases with high complexity and poor collaboration should incorporate peer-aided clinical supervision and periodic review.