K+ and glutamate released endogenous adenosine from superfused slices of rat parietal cortex. The absence of Ca2+ markedly diminished K+- but not glutamate-evoked adenosine release. Tetrodotoxin decreased K+- and glutamate-evoked adenosine release by 40 and 20%, respectively, indicating that release was mediated in part by propagated action potentials in the slices. Inhibition of ecto-5'-nucleotidase by alpha,beta-methylene ADP and GMP decreased basal release of adenosine by 40%, indicating that part of the adenosine was derived from the extracellular metabolism of released nucleotide. In contrast, inhibition of ecto-5'-nucleotidase did not affect release evoked by K+ or glutamate, suggesting that adenosine was released as such. Inhibition of glutamate uptake by dihydrokainate potentiated glutamate-evoked release of adenosine. Glutamate-evoked adenosine release was diminished 50 and 55% by the N-methyl-D-aspartate (NMDA) receptor antagonists, DL-2-amino-5-phosphonovaleric acid and (+)-5-methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5,10-imine hydrogen maleate (MK-801), respectively. The remaining release in the presence of MK-801 was diminished a further 66% by the non-NMDA receptor antagonist, 6,7-dinitroquinoxaline-2,3-dione, suggesting that both NMDA and non-NMDA receptors were involved in glutamate-evoked adenosine release. Surprisingly, K+-evoked adenosine release was also diminished about 30% by NMDA antagonists, suggesting that K+-evoked adenosine release may be partly mediated indirectly through the release of an excitatory amino acid acting at NMDA receptors.