This deservedly popular book arrived in election week. More than two UK parliaments had come and gone since its first edition – a long time in psychotherapy research? Long enough, surely, for many more psychological treatments to be rated by its authors as attaining ‘clear evidence of efficacy’? Apparently not. New admissions to this short list are few: eye movement desensitisation and reprocessing for posttraumatic stress disorder; twelve-step programmes for alcohol misuse; psychodynamic psychotherapy for borderline personality disorder and behavioural treatments for cocaine misuse. For adults, that is essentially it. Several departures balance these arrivals: all treatments for anorexia nervosa, social skills training for avoidant personality and psychosocial interventions for caregivers of the elderly. Rumours of the imminent deselection of dialectical behaviour therapy have proved unfounded, while greater volatility is evident in revisions to the list of effective treatments for children and adolescents, in line with a recent companion volume, What Works for Whom: A Critical Review of Treatments for Children and Adolescents (Reference Fonagy, Target and CottrellFonagy et al, 2002).

So much for the headlines. Without changing anything else in the text, other editorial hands might have composed different ones. As with its first edition, the real virtues of this admirable and very significant book lie elsewhere. They include unflagging lucidity when summarising and contextualising the evidence base for the various treatments (from over 2000 references); helpful considerations of the evidence for physical treatments for several common disorders; and a thorough critique of the limits, necessary and actual, of the evidence-based paradigm when applied to talking treatments. The first means that the book retains a unique place in its ability to make psychotherapy outcomes research accessible, even to those who would prefer not to read any. The second can help psychological therapists to locate and trade more profitably from their stall in the wider clinical bazaar. The last adds up to a demonstration of how formal research into efficacy and effectiveness is not ready to supplant practice-based evidence in informing decisions about treatment provision.

Some cautions are stated even more clearly now than in the previous edition; for example:

Apart from the onus this places on therapists to gather relevant evidence from local practice (and there is advice on this), the message is an important one for future mental health service users, other doctors and service commissioners to appreciate. Many psychiatrists who do not provide psychological therapies themselves are likely to need to understand the limitations as well as the strengths of the evidence-based approach on behalf of such groups.

One criticism often raised when What Works For Whom? first appeared was that, strictly, it concentrated on what (treatment) works for what (disorder). The second edition now adds a good deal more material about ‘where?’ and ‘when?’ as it addresses this question in relation to more disorders than before (notably so for substance misuse). The current state of research means that rather more can be said about all these questions for some conditions than for others, and each chapter's structure is adapted accordingly. The authors stop short of stating general recommendations for the design and publication of future research studies. If respected, such recommendations might improve studies’ compatibility and comparability and facilitate ‘meta-regression’ analyses concerning specific predictive factors. No one seems better qualified than the authors to draft such guidelines: they would make an interesting appendix to future editions. Perhaps any reticence about offering new guidance reflects the dearth of evidence that, as yet, publication of clinical guidelines has actually improved outcomes.

In clinical practice, choices ultimately involve matching person with person. Working assumptions about ‘who works for whom?’ are unavoidable. Understandably, Roth & Fonagy's comparative evaluations of the ‘whats’ remain most reticent when human equations evidently dwarf technical ones (as they do with group and non-directive therapies). However, a considerable expansion of earlier material on the impact of the therapist's training and the therapeutic alliance is supplemented by a section on ‘matching patients to therapies’. It is slender but extremely welcome, deserving expansion itself as research finds more to say about ‘whom’. This can only help the authors to justify their restated faith in formulation as the ‘critical step’ in intervention.

Already a classic, What Works For Whom is (literally) bound to stay with us. There is more substantial revision in this edition than its slightly cautious summaries betray, alongside useful innovations. No library should lack a copy. All psychiatrists should know where to find it. Psychotherapists need to read it.