Since the 19th century, we have come to think of disease in terms of specific entities--entities defined and legitimated in terms of characteristic somatic mechanisms. Since the last third of that century, we have expanded would-be disease categories to include an ever-broader variety of emotional pain, idiosyncrasy, and culturally unsettling behaviors. Psychiatry has been the residuary legatee of these developments, developments that have always been contested at the ever-shifting boundary between disease and deviance, feeling and symptom, the random and the determined, the stigmatized and the value-free. Even in our era of reductionist hopes, psychopharmaceutical practice, and corporate strategies, the legitimacy of many putative disease categories will remain contested. The use of the specific disease entity model will always be a reductionist means to achieve necessarily holistic ends, both in terms of cultural norms and the needs of suffering individuals. Bureaucratic rigidities and stakeholder conflicts structure and intensify such boundary conflicts, as do the interests and activism of an interested lay public.