The NMDA receptor-mediated component of the hippocampal granule cell population excitatory postsynaptic potential response to low frequency (< 0.2 Hz) stimulation of the medial perforant path was characterized in vivo. Extracellular recordings were obtained from the dentate molecular layer in anesthetized rabbits, and glutamatergic and GABAergic antagonists were applied locally by pressure ejection. To measure the NMDA-mediated component, the NMDA receptor antagonist D-5-aminophosphonovalerate (APV) was applied during the constant ejection of physiological saline, 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX), and/or bicuculline methiodide. In general agreement with the results of attempts by other investigators to identify NMDA responses in vivo, APV did not significantly reduce the response to a single stimulus impulse in the presence of saline. However, an NMDA-mediated response was revealed when alpha-amino-3-hydroxy-5-methyl-4-isoxazoleproprianate receptor-mediated current flow was eliminated by applying the non-NMDA receptor antagonist CNQX. The NMDA component was negative-going as predicted, but its duration was considerably less than indicated in other studies of the dentate in vitro. The relative magnitudes of the NMDA and non-NMDA components of the EPSP were found to vary as a function of stimulus intensity or frequency. The NMDA receptor-mediated component represented 12% of the control response and increased to over 25% in response to higher stimulus intensities. A brief, high-frequency burst of impulses evoked a larger NMDA component in the presence of CNQX and was able to evoke an NMDA component in the presence of saline. Surprisingly, short trains of stimulation at lower frequencies typically produced suppression of the NMDA component. In a final series of experiments, it was found that many characteristics of the NMDA component were substantially altered by GABAergic inhibition. In the presence of the GABAA antagonist bicuculline, the magnitude of NMDA receptor-mediated responses was increased and their duration was greatly extended. Additionally, in the presence of bicuculline, the NMDA component facilitated markedly in response to frequencies of stimulus input > 20 Hz. These results indicate in vivo that the initiation and duration of NMDA current flow depend strongly upon the intensity and frequency of perforant path stimulation. In addition, the NMDA response to a single impulse appears to be reduced and truncated by input from GABAA receptor-mediated feedback and/or feedforward inhibition, and this inhibition affects temporal summation of NMDA receptor-mediated responses over a wide range of input frequencies. It is suggested that such inhibition results from the activation of GABAA receptors located on granule cell dendritic shafts.(ABSTRACT TRUNCATED AT 400 WORDS)