We studied the changes in plasma arginine vasopressin in 5 patients with diabetic ketoacidosis and one patient with non-ketotic hyperosmolar coma who had marked hyperglycemia (36.6 \ m=+-\ 4.6 mmol/l, mean \ m=+-\ sem) and dehydration. Plasma osmolality (Posm) was 342.2 \ m=+-\ 11.4 mOsm/kg H2O, and hematocrit, serum protein, and blood urea nitrogen were also elevated at hospitalization. Circulating blood volume was decreased by approximately 21% as compared with that on day 7. Plasma AVP level was increased to 8.5 \m=+-\1.6 pmol/l at hospitalization. When hyperglycemia was improved by iv infusion of a small dose of insulin plus fluid administration, plasma AVP level promptly decreased to 2.4 \m=+-\0.4 pmol/l within six hours. When plasma AVP level had normalized, Posm was still as high as 305 mOsm/kg H2O, but the loss of circulating blood volume was only 4.2% of the control state. Plasma AVP level was positively correlated with change in hematocrit (plasma AVP = 3.58 + 0.45 \m=.\ hematocrit, r = 0.468, p < 0.01), serum protein (r = 0.487, p < 0.01), Posm (r = 0.388, p < 0.01), and blood glucose (r = 0.582, p < 0.01). Plasma AVP level was negatively correlated with the change in circulating blood volume (plasma AVP = 3.6 \p=n-\0.14 \ m=. \ change in circulating blood volume, r = \m=-\0.469, p <0.01). These results indicate that both non-osmotic and osmotic stimuli are involved in the mechanism for AVP release in patients with diabetic coma, and that the non-osmotic control of AVP may contribute to circulating homeostasis, protecting against severe blood volume depletion in diabetic patients suffering from hyperglycemia and dehydration.We have demonstrated that plasma arginine vaso¬ pressin increases in patients with non-insulin-de¬ pendent diabetes mellitus whose fasting blood glu¬ cose is moderately elevated (9.5-10.1 mmol/1, mean value) (1). Such an increase in plasma AVP normalized at the time of discharge, and it is closely related to the decrement in circulating blood volume.Marked elevation of blood glucose is common in ketotic and non-ketotic diabetic coma, accompa¬ nied by marked glucosuria and osmotic diuresis. The patients are thus suffering from serious dehy¬ dration. Such volume-depleted states usually in¬ crease plasma renin activity (PRA) and plasma aldosterone concentration (2). This has been ex¬ plained by the decrease in circulating blood volume, mediated through an activation of baroreceptors. Several reports have shown that plasma AVP is also elevated in diabetic ketoacidosis (3-6).Zerbe et al. noted the relative roles of osmotic and non-osmotic stimuli in AVP response (3). The phys¬ iological significance of elevated AVP and the mechanism for the increase in AVP release in con¬ ditions such as ketoacidosis are still unclear, how¬ ever.In the present study we determined whether plasma AVP increases in diabetic coma, and what mechanisms are involved in the increase in AVP release. Also, we investigated how fast the in¬ creased level of plasma AVP returns to the normal range after the start of in...