Random estimations of plasma arginine vasopressin concentration were undertaken in 6 non-oedematous patients receiving diuretic therapy for hypertension, who were admitted to hospital with severe hyponatraemia. Hyponatraemia resolved within 2 weeks of discontinuing the diuretic. Measurable amounts of plasma arginine vasopressin were detected in all 6 patients. Sequential biochemical measurements in one patient, performed when plasma sodium concentration and osmolality were returning to the normal range, disclosed that urine osmolality remained higher than plasma osmolality during the first 5 days, when urine volume and sodium excretion were low. Thus the rise in plasma sodium was not initially related to water diuresis. The ability to excrete a water load was severely limited on the fifth day, but improved progressively by the tenth and seventeenth days. Diuretic-induced hyponatraemia is associated with incomplete suppression of anti-diuretic hormone secretion arising from non-osmotic stimulation, in conjunction with transient impairment of renal diluting ability which could be due to net sodium deficit.