Research requires reduction of complex phenomena to more basic ones. The degree of reduction seen in Behaviour Research and Therapy (BRAT) papers in general, and of Rachman's work in particular, meets with two types of criticism. First, it is argued that reduction is taken too far. This objection is typically heard from critics challenging the generalisability of outcome trials. Ignoring between-patient differences in protocollisation of treatments in outcome studies is held to provide an underestimation of what individualised therapy has to offer. Alternatively, the strictness of inclusion/exclusion criteria in trials is held to give an overestimation of treatment effects in ordinary clinical practice. Scholars studying information processing or biological psychology/psychiatry argue that reduction has not been taken far enough. Convictions, beliefs, schemas, etc. are held not to explain behaviour or emotion. They should be reduced to perculiarities in information processing or, according to others, to neurophysiological processes. The objection that the reduction inherent in clinical trials hampers generalisation is obviously an empirical issue. It is argued that the objection that reduction has not been taken far enough is also an empirical issue. Given the present state of knowledge in psychopathology, and believing that degrees of reduction is an empirical rather than a philosophical issue, the conclusion is that both objections are empirically invalid and that the degree of reduction in Rachman's work has proved balanced and productive.