Anorexia nervosa may sometimes be resistant to all forms of therapy, with cases running through a gamut of somatic and psychological treatments. One possible explanation for this may be that the illness is of heterogeneous aetiology (Kay and Leigh, 1938; King, 1963; Feighner et al, 1972), although others regard anorexia as a single condition that tends to breed true though having protean manifestations (Russell, 1970; Crisp et al, 1980). In most of the larger series of cases, a proportion have depressive symptoms both during the illness (Dally, 1969; Theander, 1970; Crisp et al, 1980) and some years later (Morgan and Russell, 1975; Hsu et al, 1979), and a family history of affective disorder is also commonly reported (Dally, 1969; Theander, 1970; Morgan and Russell, 1975). Cantwell et al (1977) have reviewed the complex relationship between depression and anorexia nervosa and have suggested that some of the remedies used in the treatment of depression merit further exploration in the management of anorexia nervosa. We report here a patient who was in hospital for over four years and in whose eventual improvement lithium may have played an important role.