The role of arginine vasopressin (AVP) in the regulation of renal medullary blood flow is uncertain. To determine if AVP has a direct vasoconstrictive action on vasa recta, the effect of AVP on erythrocyte velocity (VRBC), diameter, and blood flow (QVR) in descending vasa recta (DVR) and ascending vasa recta (AVR) was studied in the exposed renal papilla of four groups of chronically water diuretic rats using fluorescence videomicroscopy. There were three periods: control (period 1), experimental (period 2), and recovery (period 3). In periods 1 and 3, all groups received hypotonic saline. In period 2, group I rats (AVP) received AVP (45 ng/h per kg body wt); group II (time) received hypotonic saline alone; group III (AVP plus V1-inhibitor) received AVP plus its vascular antagonist, d(CH2)5Tyr(Me)AVP; and group IV (V1-inhibitor) received the vascular antagonist alone. Another group of rats (group V) was employed to demonstrate that the rise in blood pressure induced by a 3- or 10-ng/kg injection of AVP was virtually abolished by the prior infusion of the V1-inhibitor. The urine of group III as well as group I rats was concentrated (Uosm = 721 +/- 62 H2O vs. 670 +/- 39 mosM/kg), while urine remained dilute in groups II and IV. In period 2, VRBC and QVR in DVR and AVR decreased in group I, did not decrease in group III, and increased in groups II and IV. The vascular antagonist thus completely abolished the AVP-induced decrease in QVR in group III. These findings unequivocally establish that AVP in physiological amounts reduces medullary blood flow, at least in part, by a direct vasoconstrictive action on the medullary microcirculation. They also show that an effect of AVP on medullary blood flow is not necessary for its antidiuretic effect.