Gaudiano's review (this issue) raises fundamental issues about the process of evaluating a new psychological therapy for psychosis. Psychosis has previously been seen as intractable and psychological interventions as likely to make things worse. As a result, studies are faced not only with the normal difficulties of demonstrating methodological rigor but also with extra ones such as professional disbelief. Further, there are almost no equivalent, evidence-based psychological therapies to compare it with; focusing on “key” interventions is unlikely to be productive in such a complex disorder, and even if found useful there is the future problem of training and supervising sufficient practitioners to provide it. Developing models, devising and testing hypotheses and incorporating user views are advocated.