Abstract Opioid actions on oxytocin secretion into blood and cerebrospinal fluid (CSF) were investigated in urethane-anaesthetized female rats after intracerebroventricular (icv) infusion of morphine sulphate or vehicle for 5 days. Serial femoral arterial blood samples and cisterna magna CSF samples were collected for radioimmunoassay. Naloxone was given to assess endogenous opioid tone in icv vehicle-infused rats and to precipitate withdrawal in morphine-dependent animals. Initial plasma oxytocin concentration was not affected by icv morphine infusion. In control rats receiving icv vehicle, naloxone increased plasma oxytocin 11-fold within 5 min, and in icv morphine-infused rats, naloxone increased plasma oxytocin 80-fold within 5 min. In both groups, 90 min after naloxone plasma oxytocin was still 5 and 10 times, respectively, the initial concentration. Without naloxone, neither plasma nor CSF oxytocin concentration changed significantly with time (up to 90 min) in either icv treatment group. In the icv vehicle group, there was a 2-fold increase in CSF oxytocin 90 min after naloxone. In the icv morphine-infused group, CSF oxytocin was increased 5-fold 40 min after naloxone. In another group of icv morphine-infused rats, intravenous infusion of oxytocin to achieve plasma levels similar to those seen after naloxone, did not significantly increase CSF oxytocin. In a further group of icv morphine-infused rats, [(3)H]oxytocin was infused intravenously immediately after naloxone was given; in these rats oxytocin transfer from blood to CSF could account at most for only 20% of the increase in CSF oxytocin after naloxone. A further group of rats underwent bilateral microknife ablation of the paraventricular nuclei (PVN) 9 days before icv vehicle or morphine infusions were started; blood and CSF samples were collected under urethane anaesthesia. Initial concentrations of oxytocin in CSF and in plasma were similar in both groups with PVN ablation. In all PVN-lesioned rats initial plasma concentrations of oxytocin were undetectable (<5 pg/ml) and thus less than in intact rats. In contrast, initial levels of oxytocin in CSF were 8-fold greater in PVN-lesioned rats than in intact animals. Naloxone increased plasma oxytocin concentration in the icv vehicle group at least 10-fold within 30 min and in the icv morphine group at least 100-fold within 5 min. CSF oxytocin in the icv vehicle group was not altered by naloxone, but in the icv morphine group CSF oxytocin was increased 5-fold 40 min after naloxone. There were no consistent differences between the icv vehicle- and icv morphine-treated groups in the initial plasma levels of vasopressin, growth hormone and adrenocorticotrophin; PVN ablation did not affect adrenocorticotrophin levels. After naloxone growth hormone levels did not change, vasopressin concentration rose moderately only after 90 min and only in the icv vehicle-treated group, and adrenocorticotrophin concentrations decreased with time whether or not naloxone was given. The results imply an endogenous opioid tone on neurons releasing oxytocin into CSF, and morphine-dependence of these neurons. Furthermore, in PVN-lesioned rats, magnocellular supraoptic neurons could be a source of oxytocin release into CSF.