Antidromically identified paraventricular neurones were recorded simultaneously with intramammary pressure in urethane (1.2 g/kg) anaesthetized rats during suckling. The correlation of the firing pattern of these neurones with milk ejection enabled distinction between oxytocin and vasopressin neurones. Oxytocin neurones displayed a short (2-6 s) characteristic high-frequency burst of spikes. This activation probably occurred simultaneously in all oxytocin neurones 12-18 s before milk ejection and was regular in both frequency and amplitude (total number of spikes). The role of neurohypophysial peptides and analogues in the control of these characteristics was studied. Injecting 10 pg, 100 pg and 1 ng of oxytocin into the 3rd ventricle increased background activity of slow-firing oxytocin neurones (less than 3 spikes/s) and had a strong dose-dependent facilitatory effect on the milk ejection reflex, increasing both the amplitude and frequency of neurosecretory bursts. No effect was observed on non-neurosecretory neurones. Such injection also triggered the milk ejection reflex when it had not appeared an hour after suckling began. Oxytocin did not itself induce neurosecretory activation, which only appeared if the young rats were sucking. Injecting oxytocin into the lateral ventricle was less effective than into the 3rd ventricle. No effect was observed after injection into the venous blood or into the 4th ventricle, which suggested that oxytocin acts in the hypothalamus. Injecting mesotocin or isotocin into the 3rd ventricle had a facilitatory effect similar to that of oxytocin but vasopressin, vasotocin, MIF I (pro-leu-gly-NH2, terminal triplet oxytocin) or bovine neurophysins I and II did not modify neurosecretory activation or the milk ejection pattern. Injecting an oxytocin antagonist, ([1(beta-mercapto-beta, beta cyclopentamethylene propionic acid), 8-ornithine] vasotocin, d(CH2)5OVT) into the 3rd ventricle decreased milk ejection frequency and considerably delayed the reappearance of the first milk ejection. This resulted from a decrease in both frequency and amplitude of neurosecretory bursts, which were too small to induce detectable oxytocin release. Moreover, d(CH2)5OVT suppressed the facilitatory effect of exogenous oxytocin. Under normal conditions, endogenous oxytocin seemed to be involved in the control of neurosecretory activation. Injecting 1 ng oxytocin or 1 or 10 ng vasopressin into the 3rd ventricle did not modify the firing pattern of vasopressin neurones whether activated by hyperosmotic stimulation (1 ml NaCl, 9% solution (w/v) I.P.) or not.(ABSTRACT TRUNCATED AT 400 WORDS)