This article argues that the current approach to guideline development for the treatment of depression is not supported by the evidence: clearly depression is not a disease for which treatment efficacy is best determined by shortterm randomised controlled trials. As a result, important findings have been marginalised. Different principles of evidencegathering are described. When a wider range of the available evidence is critically considered the case for dynamic approaches to the treatment of depression can be seen to be stronger than is often thought. Broadly, the benefits of shortterm psychodynamic therapies are equivalent in size to the effects of antidepressants and cognitive-behavioural therapy (CBT). The benefits of CBT may occur more quickly, but those of shortterm psychodynamic therapies may continue to increase after treatment. There may be a ceiling on the effects of shortterm treatments of whatever type. Longerterm psychodynamic treatments may improve associated social, work and personal dysfunctions as well as reductions in depressive symptoms.David Taylor, a consultant psychotherapist at the Tavistock & Portman NHS Foundation Trust (120 Belsize Lane, London NW3 5BA, UK. Email: dtaylor@taviport.nhs.uk), is the clinical lead of the Tavistock Adult Depression Study (a randomised controlled trial of 60 sessions of weekly psychoanalytic psychotherapy v. treatment as usual for patients with chronic, refractory depression). He is a training and supervising psychoanalyst at the Institute of Psychoanalysis.