The diabetogenic action of pituitary extracts containing growth hormone has been recognised for more than 60 years and the importance of growth hormone in the development and progression of diabetic retinopathy for more than 30 years. Hypophysectomy was the first effective treatment for retinopathy but was discontinued because of the risk of severe hypoglycaemia that it produced and the development of an alternative, less dangerous therapy--photocoagulation. The precise role and significance of growth hormone in diabetes care, however, remains to this day a mystery. The fact that modern, highly purified biosynthetic preparations of growth hormone still retain full diabetogenic potency and the fact that diabetes develops in up to 25% of patients with acromegaly indicate growth hormone's potential for involvement in the aetiology of diabetes mellitus, although most will agree that this is not likely to be an important factor in the large majority of 'idiopathic' cases. There is strong evidence to indicate a substantial hypersecretion of growth hormone in 'idiopathic' diabetes mellitus (particularly insulin-dependent cases and those with retinopathy), which appears to be more related to residual pancreatic insulin secretion than to metabolic control. Since the advent of biosynthetic growth hormone in sufficient quantity to perform trials in adults, we are more aware of growth hormone's considerable potency in the regulation of body composition, growth factor production and intermediary metabolism. In this article, we review the literature and, from this and our own work, propose a new hypothesis which links the hypersecretion of growth hormone to reduced hepatic secretion of insulin-like growth-factor I (IGF-I) as a direct result of reduced portal insulin levels in diabetes mellitus. The hypersecretion of growth hormone exposes peripheral organs such as the retina and kidney to conditions favouring the expression of growth-hormone-dependent growth factors such as IGF-I which may contribute to the development of diabetic microvascular disease by autocrine and/or paracrine effects. If this hypothesis proves to be true, it offers new opportunities for the prevention of diabetic microvascular complications through suppression of growth hormone secretion which in turn will increase insulin sensitivity and facilitate good glycaemic control.