1. We measured plasma concentrations of arginine vasopressin (AVP), arterial, venous and amniotic pressures, and carotid and femoral blood flows in fifteen chronically instrumented fetal sheep at 119-125 days of gestation. In eight of the fetuses the carotid sinus nerves were cut (denervated fetuses); the other seven remained intact and served as controls (intact fetuses). 2. In the intact fetuses during hypoxia there was an increase in plasma [AVP] and in perfusion (arterial-venous) pressure, a transient bradycardia, and an increase in carotid and a decrease in femoral blood flow. Whilst femoral vascular resistance (perfusion pressure/femoral blood flow) increased, there were no changes in carotid vascular resistance during hypoxia. 3. In the denervated fetuses no significant bradycardia, fall in femoral blood flow or increase in femoral vascular resistance was present soon after the onset of hypoxia but plasma AVP increased to similar concentrations to those observed in intact fetuses during hypoxia. 4. We conclude that carotid denervation does not affect plasma [AVP] during hypoxia in fetal sheep. This suggests that (1) AVP release during hypoxia is not mediated by a carotid chemoreflex and (2) AVP does not play an important role in these initial fetal cardiovascular responses. Furthermore, we previously reported that intact fetuses survive acute hypoxia better than denervated fetuses following phentolamine treatment, and we believe this to be due to the action of a non-alpha-adrenergic vasoconstrictor released in part via a carotid chemoreflex. The present results suggest that this vasoconstrictor is not AVP.