1. The release of endogenous GABA and glutamate from the cerebral cortex was measured using a cortical cup technique in unanaesthetized freely moving rats and anaesthetized rats by means of a sensitive and specific mass-spectrometric procedure. 2. GABA release was not affected by the presence of the dura mater or by anaesthesia. Glutamate output was reduced by urethane but not by pentobarbital anaesthesia and by the presence of the dura. 3. An isotonic solution containing 50 mM KCl placed epidurally within the cup elicited a significant short-lasting increase in glutamate output, a decrease in GABA output and a short-lasting electrocorticogram (ECoG) activation. 4. When the dura was removed, a high K+ solution placed on the exposed cerebral cortex elicited a 7--8 fold increase in GABA output accompanied by a marked decrease in glutamate output and by ECoG synchronization. The changes in GABA and glutamate output had parallel time-course and were prevented by the application within the cup of tetrodotoxin (3 X 10(-5) M). 5. Amphetamine at the doses of 3.7 and 7.4 mumol . kg-1 i.v. increased glutamate output and at the dose of 37 mumol . kg-1 i.v. increased GABA output. Both effects were prevented or reduced by haloperidol pretreatment (0.65 mumol . kg-1 i.v.). 6. It is concluded that GABA and glutamate released from the cerebral cortex and diffused into an epidural or cortical cup originate at least in part from the brain. The rate of their release is influenced by changes in neuronal activity. The measurement of their rate of release offers a useful tool for the study of the functional role of cortical GABA and glutamate-releasing neurons.