1 Roth et al. (1972) have proposed that within the affective disorders there are two distinct syndromes--anxiety and depression--differentiated by the relative predominance of these symptoms, the presence or absence of premorbid maladaptive traits, and the natural outcome of the disorder. 2 Goldberg (1982) on the other hand has suggested that there is a substantial overlap between the two syndromes, and the symptoms of anxiety and depression correlate highly with each other. 3 Patients tend to be treated with an anxiolytic or an antidepressant drug according to the predominance of anxiety or depressive symptoms. 4 There is conflicting evidence regarding the relative effectiveness of tricyclic antidepressants in psychotic (or endogenous) and neurotic (or reactive) depressive illness. Paykel (1972), however, found that the psychotic-neurotic dichotomy was not helpful in predicting outcome, whereas anxious depressives tended to respond poorly. 5 In three patients with affective illness, tested at regular intervals over the course of several weeks, anxiety and depressive symptoms fluctuated in a way which suggested that they were yoked to each other. A worsening indicated simultaneous deterioration of these symptoms. When they improved, this may have been due to the beneficial effects of the antidepressant drug administered. These findings suggest that antidepressants act, or fail to act, in a fundamental manner on the illness itself rather than on individual symptoms.