In 1934 the Chicago physician R.T. Woodyatt suggested that 'The history of diabetes has been marked by recurrence of certain ideas which decline and disappear; only to go through a similar cycle again in an altered form in a new generation'. This has been particularly true of the concept of brittle diabetes which Woodyatt himself introduced in the 1930s. He never wrote a paper on the subject but contemporaries understood it to refer to excessive fluctuations of blood sugar which could not be explained by patient or physician errors; the cardinal feature was unpredictability and unexpected hypoglycaemic reactions. Also in the 1930s, practitioners of the newly formed psychosomatic movement took an interest in the effect of emotional factors on the course of diabetes and, in particular, patients who were 'difficult' or 'refractory'. What marked 'difficult' patients was that they did not follow their doctor's instructions or had recurrent diabetic ketoacidosis. By the 1950s the question was whether there were two distinct groups of patients; one whose lability could be cured by adjusting insulin, diet, and exercise, and another whose lability had an emotional origin. Did proponents of the organic school have patients (unreported) in whom lability had an obvious emotional cause or, conversely, were the psychosocial problems which the psychiatrists unearthed a consequence rather than a cause of the instability? My experience with a patient with factitious hypoglycaemia which remained undetected for weeks in a clinical research unit suggested that neither close observation nor screening by a psychiatrist could rule out factitious disease. Therefore in 1977 I suggested that the definition of brittle diabetes should be a patient whose life was 'constantly disrupted by episodes of hypo- or hyperglycaemia, whatever their cause'. This was widely accepted and there was a subtle shift towards regarding brittle diabetes as synonymous with recurrent ketoacidosis. In the 1980s two English and one American group investigated large series of such patients, using new methods to try to uncover a biochemical basis such as defective insulin absorption, accelerated degradation at insulin injection sites, and inappropriate secretion of various counterregulatory hormones. Most of these patients were young overweight women and the eventual conclusion was that in most the instability was self-induced. In the 1980s recurrent, often warningless, hypoglycaemia was recognized as a problem in its own right but in this new generation was reborn as a problem of insulin pharmacokinetics as Woodyatt originally conceived it.