1. Indirect evidence suggests that the concentration of arginine vasopressin in the plasma of normally hydrated man is about 1 mu-u./ml., but this is usually considered to be below the limit of sensitivity of the standard assay preparation, the water-loaded Wistar rat under ethanol anaesthesia.2. It was found that there was a surprising variation in sensitivity to vasopressin between batches of Wistar rats, and that other varieties of rat (including those with diabetes insipidus) were no more sensitive.3. Three modifications of the standard assay procedure produced an increase in sensitivity:(a) using Wistar rats weighing 100-150 g, rather than larger animals;(b) commencing the assay shortly after surgery, i.e. as soon as the urine flow reached 25 mul./min;(c) infusing vasopressin intravenously (0.5-3 mu-u./min). By using modification (a) with either (b) or (c) it was possible to detect as little as 0.5 mu-u.4. With these modifications antidiuretic activity equivalent to 0.5-2.0 mu-u./ml. of arginine vasopressin was measured in nine samples of plasma from a normally hydrated subject.5. It is suggested that the frequent reports of enhanced sensitivity may have been due to the fortuitous use of a particularly sensitive batch of rats, or to a high endogenous secretion of vasopressin due to operative trauma.