The idea that power relations structure social life is self-evident to most anthropologists. Western medical knowledge or biomedicine, and by extension science or scientific knowledge, however, has until relatively recently been exempt from anthropological scrutiny in political terms. An understanding of biomedicine as a system of knowledge that is not a copy of facts but a representation of them has entailed a break with the traditional separation of folk knowledge and scientific knowledge in anthropology, making it possible to include biomedicine in the repertoire of ethnographic objects. The pecu- liarity of biomedicine as a cultural system, seen from this perspective, lies in a paradox: its self-characterization as a set of non-ideological discourses and practices is a representation that conceals its ideological and power-saturated nature. Through an analysis of DSM-IV-TR, this article explores some of the representational strategies through which this con- cealment takes place in biomedical psychiatry: the asocial and universal character of mental illness categories; the neu- trality of clinical practice; and the non-moral nature of clinical criteria and judgment. These are concealed metaphors in the true sense, for not only do they speak of something without naming it but they also deny their own existence as metaphors.