N-Methyl-D-aspartate (NMDA) receptor dose-response relationships that are based on macroscopic currents suggest that NMDA and a different agonist molecule, glycine, must together activate the channel. Since single-channel recordings have a much higher resolution than whole-cell currents, they provide a highly sensitive test for the absolute requirement of NMDA channel opening for glycine. Rapid application of 10-300 microM NMDA to outside-out patches from cultured cortical neurons evoked substantial single-channel activity in the absence of added glycine. However, in the presence of a high affinity and highly selective glycine-site antagonist, 5,7-dichlorokynurenate (DCK), NMDA failed to evoke any openings on its own. Channel openings could not be produced by saturating concentrations of NMDA (up to 1 mM) but were evoked when glycine was added to the test solution. Glycine alone (up to 100 microM) was similarly ineffective in the continuous presence of D(-)-2-amino-5-phosphonovaleric acid (D-APV), an NMDA-site antagonist. Reversal of antagonist blockade by the appropriate ligand (glycine or NMDA) and the normal appearance and duration of channel openings evoked in the presence of either antagonist ruled out open channel block. These single-channel data confirm the hypothesis that both NMDA and glycine are coagonists of the NMDA receptor. Furthermore, the coagonist requirement increases the potential targets for therapeutic drugs aimed at blocking the pathologies resulting from overactivation of NMDA receptors.