This chapter focuses particularly on the use of randomized controlled trials (RCT) to assess treatment efficacy. In the second part it diagnoses what has gone wrong with the use of RCTs in psychiatry. Richard Ashcroft argues that RCT methodology implicitly commits one to various assumptions about the nature of probability judgements. There has been a great deal of discussion of the methodological and ethical problems associated with the use of RCTs: many authors have worried that an emphasis on RCTs has led to other types of evidence such as the individual case study being undervalued, Williams & Garner 2002. The chapter argues the case of RCTs in psychiatry shows that traditional epistemology and philosophy of science have taken too narrow a focus when thinking about scientific method. It concludes that if epistemologists remember that scientists are real people rather than ideal truth-seekers then this will make a difference to the kinds of methods they will recommend.