I am writing to comment on this debate (Reference Turkington and McKennaTurkington & McKenna, 2003) from the point of view of a practising clinician who regularly works with people who are experiencing or recovering from psychotic symptoms. I find both cognitive–behavioural and psychodynamic models useful in my work.

Many research practitioners have reservations about using randomised controlled trials as the main test of the effectiveness of psychotherapeutic interventions (see Reference McPherson, Richardson and LerouxMcPherson et al, 2003). Problems raised relate to use of a method developed for trials of medication on the assumption that psychotherapy works in the same way. For example, blindness to modality in patients and therapists is not possible, randomisation is ethically questionable and is unlikely to bring about the best clinical outcomes, and use of control groups is problematic as it is impossible to control the variables sufficiently to be sure of revealing specific effects. Given these reservations, the apparently small effects of cognitive–behavioural therapy (CBT) with psychosis obtained in randomised controlled trials might be seen as very encouraging. Many studies have included medication-resistant patients, which adds to their significance.

It is interesting that some of the controls employed, such as supportive counselling and befriending, also obtained short-term improvements over treatment as usual. Much more work needs to be done to tease out the different active ingredients in different kinds of work with individuals. This was the conclusion reached by a detailed meta-analysis of CBT (and other interventions) for schizophrenia (Reference Pilling, Bebbington and KuipersPilling et al, 2002). There is recent evidence for effectiveness of psychodynamic psychotherapy with psychosis under certain conditions (Reference JacksonJackson, 2001). Cognitive and psychodynamic psychotherapists have begun to explore areas of common ground, possibilities for recognising the different contributions from different approaches to therapeutic work and the issue of suitability of individuals to different paradigms of intervention (Reference MiltonMilton, 2001). Furthermore, there is currently much interest in the contribution of the social environment to ongoing disability in psychosis, which may link with the success of befriending. The Department of Health's (2001) Mental Health Policy Implementation Guide on early intervention in psychosis encourages services to address issues around stigmatisation and social marginalisation. These areas of intervention can combine in a flexible and holistic approach that is both sophisticated and acceptable to individuals with psychosis, and that (along with the undoubted contribution of medication) offers them worthwhile options for treatment and support into recovery.