This review attempts to outline the present understanding of diabetic autonomic neuropathy. The clinical features have been increasinly recognised but knowledge of the localization and morphology of the lesions and their pathogenesis remains fragmentary. A metabolic causation as postulated in somatic nerves accords best with clinical observations. Most bodily systems, particularly the cardiovascular, gastrointestinal and urogenital, are involved with added disturbances of thermoregulatory function and pupillary reflexes. Possible effects on neuroendocrine and peptidergic secretion and respiratory control await definition. Current interest centres around the development of a new generation of tests of autonomic nerve function that are simple, non-invasive, reproducible and allow precision in diagnosis and accurate quantitation. Most are based on cardiovascular reflexes and abnormality in them is assumed to reflect autonomic damage elsewhere. Probably no single test suffices and a battery of tests reflecting both parasympathetic and sympathetic function is preferable. Little is known of the natural history. The prevalence may be greater than previously suspected and although symptoms are mild in the majority, a few develop florid features. The relation of control and duration of diabetes to the onset and progression of autonomic neuropathy is not clearly established. Once tests of autonomic function become abnormal they usually remain abnormal. Symptomatic autonomic neuropathy carries a greatly increased mortality rate possibly due to indirect mechanisms such as renal failure and direct mechanisms such as cardio-resiratory arrest. Improved treatment of some of the more disabling symptoms has been possible in recent years.