Evidence-based psychiatry (EBP) has arisen through the application of evidence-based medicine (EBM) to psychiatry. However, there may be aspects of psychiatric disorders and treatments that do not conform well to the assumptions of EBM. This paper reviews the ongoing debate about evidence-based psychiatry and investigates the applicability, to psychiatry, of two basic methodological features of EBM: prognostic homogeneity of clinical trial groups and quantification of trial outcomes. This paper argues that EBM may not be the best way to pursue psychiatric knowledge given the particular features of psychiatric disorders and their treatments. As a result, psychiatry may have to develop its own standards for rigour and validity. This paper concludes that EBM has had a powerful influence on how psychiatry investigates and understands mental disorders. Psychiatry could influence EBM in return, reshaping it in ways that are more clinically useful and congruent with patients' needs.