Objective: To detail limitations to level I evidence derived from randomised controlled trials of antidepressant treatments and which is held to be fundamental to the development and validity of treatment guidelines.
Method: Recent efficacy studies and meta-analyses of treatments of major depression are considered.
Results: The largest database in psychiatry--demonstrating that all principal treatments are of similar efficacy, and that antidepressant drugs are not distinctly superior to placebo treatment--is unlikely to be valid.
Conclusion: Excessive belief in and weighting of the evidence emerging from randomised controlled trials deserves to be criticized. An argument is put for adopting alternative approaches to evaluating the likely effectiveness of any antidepressant treatment.