Sylvia Plath, a gifted young poet, died by her own hand in London forty years ago. Subsequently a friend and fellow-poet, Al Alvarez, included a personal account of her final illness, as well as of his own unconnected suicidal attempt, in The Savage God: a Study of Suicide. 1 A review in the British Journal of Psychiatry by Eliot Slater, at that time its editor, concluded that Alvarez had failed to grasp the essential difference between his own depression and hers: ‘His own experience’, he wrote, ‘so similar as he must think, does not help him to understand the act of Sylvia Plath; he is cut off by the gulf that separates the reactive from the endogenous depression, alike in appearance but different in kind. Sylvia Plath was a manic-depressive. ’2 How well do these comments stand up today? Leave aside the fact that the two suicidal acts were not so different in terms purely of dangerousness, and that in each instance chance played some part in deciding on survival or death. The question remains of interest because it throws light on changing trends in the nosology and classification of psychiatric disorders, as well as on the psychology of a writer who since her untimely death has acquired iconic status both for the feminist movement and for a new literary generation. Dr Slater's appraisal was based on a binary model of affective illness, in which the terms endogenous and reactive—or alternatively psychotic and neurotic—were used to denote two contrasting syndromes: the one consisting of recurrent severe disorders of mood (depressive and/or manic), apparently spontaneous in onset, the depressive phase being accompanied by psychomotor retardation, feelings of guilt and unworthiness, early-morning waking and somatic changes; the other presenting as milder, often intermittent depression mingled with anxiety, triggered by adverse life events, marked by irritability and self-concern rather than guilt and by subjective complaints rather than objective bodily disorder. 3 If one accepts this model, there is undoubtedly a strong case for allocating Sylvia to the former group. It was given support by John Horder, the respected London general practitioner (GP) who treated her in her last illness and who later wrote: ‘I believe, indeed it was repeatedly obvious to me, that she was deeply depressed, “ill”, “out of her mind”, and that any explanations of a psychological sort are inadequate... ’. 4 In the intervening decades, however, the underlying assumptions of this model have been seriously challenged. To begin with, belief in a natural boundary or discontinuity between the main types of depression has been weakened by repeated failure to confirm a bimodal distribution, or to demonstrate any point of rarity between the subgroups, when the clinical features of unselected case series are submitted to discriminant function analysis. 5,6 Secondly, there has been a corresponding failure to demonstrate clear-cut differences in the frequency of provoking stresses. Brown and his co-workers, 7 on the basis of detailed standardized assessments, found that depressed patients diagnosed as psychotic and neurotic, respectively, did not differ in experience of adverse life events over the months before illness onset. Indeed, age and experience of ‘past loss’ were the only personal characteristics that distinguished between them. When patients with and without a history of preceding severe life events were compared, only a slight tendency was found for the latter group to have more psychotic features. Thirdly, US workers, bringing together evidence from various research fields, concluded that depressive illness represents the final common pathway of several different pathogenetic processes focused on the diencephalon, and is essentially the same whether the processes in question are biochemically, experientially or behaviourally triggered. 8 These various findings, whose influence on psychiatric thinking became apparent when the standard classifications were last revised (see later), do not imply that individual cases cannot be characterized as typically endogenous or typically reactive. They do, however, suggest that the distribution as a whole corresponds to a dimensional rather than a categorical model, in which such pure-culture cases represent the extremes on a continuum and are likely to be greatly outnumbered by others presenting intermediate or mixed clinical pictures. Sylvia Plath's illness can be placed roughly on this continuum because information now in the public domain, including her posthumously published journals9 and letters home, 10 a semi-autobiographical novel, 11 the memoirs of family and friends, a well-researched, sensitive biography12 (one of several) and a life of her husband, 13 together provide material for a reconstruction of the case history.