A radioimmunoassay has been developed that permits reliable measurements of plasma arginine vasopressin (AVP) at concentrations as low as 0.5 pg/ml in sample volumes of 1 ml or less. Nonhormonal immunoreactivity associated with the plasma proteins is eliminated by acetone precipitation before assay, leaving unaltered a component that is immunologically and chromatographically indistinguishable from standard AVP. Storage of plasma results in a decline in AVP concentration and, thus, must be carefully regulated. The plasma AVP values obtained by our method approximate the anticipated levels and vary in accordance with physiologic expections. In recumbent normal subjects, plasma AVP ranged from (mean +/-SD) 5.4+/-3.4 pg/ml after fluid deprivation to 1.4+/-0.8 pg/ml after water loading, and correlated significantly with both plasma osmolality (r=0.52; P<0.001) and urine osmolality (r=0.77; P<0.001). After fluid restriction, plasma AVP was uniformly normal relative to plasma osmolality in patients with nephrogenic diabetes insipidus and primary polydipsia but was distinctly subnormal in all patients with pituitary diabetes insipidus. The infusion of physiologic amounts of posterior pituitary extract caused a dose-related rise in plasma vasopressin that afterwards declined at the expected rate (t(1/2)=22.5+/-4 min). We conclude that, when used appropriately, our radioimmunoassay method provides a useful way of assessing AVP function in man.