Abstract Morphological and pharmacological evidence suggest that the dense GABAergic innervation of the supraoptic nucleus is important for regulating the electrical activity of vasopressin and oxytocin neurons. We have employed the technique of intracranial microdialysis to examine extracellular GABA concentrations in the supraoptic nucleus of the anaesthetized rat and questioned whether differences exist in the dynamics of GABA release between virgin and lactating rats, and if events during lactation or following blood pressure manipulation alter endogenous GABA levels in this nucleus. No significant differences were detected between virgin and lactating animals in either basal or 100 mM potassium ion-evoked GABA release. The inclusion of the GABA uptake blocker nipecotic acid (0.5 mM) into the dialysate resulted in a six- to eight-fold increase ( P < 0.01) in GABA outflow in both groups of animals. In lactating rats, GABA outflow measured at 4 min intervals was not altered during a 60 min period of suckling by a full litter of pups and no significant change in GABA outflow was detected in relation to individual milk ejections. In virgin rats, removal of 1.5–2 ml of blood resulted in a 30–60 mmHg fall in blood pressure and a non-significant decline in GABA outflow. Replacement of blood resulted in an abrupt 50 mmHg increase in blood pressure and a significant 22% increase in GABA outflow ( P < 0.01), but no change in aspartate or methionine concentrations. Repeated intravenous injections of the α-adrenoceptor agonist, metaraminol, similarly evoked approximately 50 mmHg increments in blood pressure and a 26% increase in GABA outflow ( P < 0.05). Electrical stimulation of the diagonal band of Broca for 10 min produced a two-fold increase in GABA outflow from the supraoptic nucleus ( P < 0.05). These results show that the overall profile of basal and potassium-stimulated GABA concentrations in the supraoptic nucleus is not substantially different between lactating and virgin rats. In lactating animals we have found that GABA levels are not altered in response to suckling or at the time of high-frequency firing by oxytocin neurons to induce milk ejection. In contrast, our data further support the hypothesis that GABA inputs to supraoptic neurons are part of a baroreceptor reflex, relaying through the diagonal band of Broca, to signal periods of acute hypertension and inhibit the firing of vasopressin neurons. Such observations suggest the physiological importance of GABA inputs to the supraoptic nuclei and indicate that GABA may be used in a stimulus-specific manner to influence the activity of magnocellular neurons.