1. Ca2+-dependent inactivation (CDI) of NMDA receptors was studied using both cultured embryonic rat dorsal horn neurons and acutely dissociated postnatal rat dorsal horn neurons. The perforated patch recording method was employed in order to preserve intracellular Ca2+ buffers and other cellular constituents. In this way, the kinetics of intracellular Ca2+ concentration ([Ca2+](i)) transients and other second messenger signalling systems were maintained in a relatively normal condition. 2. Continuous application of 30 μM NMDA to cultured dorsal horn neurons voltage clamped at -70 mV evoked currents that inactivated to about 40% of the peak value with time constants between 200 and 600 ms. CDI with similar kinetics was also observed in acutely dissociated postnatal rat dorsal horn neurons. 3. When NMDA was applied in a low (20 μM) Ca2+ bath or when dorsal horn neurons were held at +70 mV, inactivation was either very weak or absent. The peaks of NMDA currents were significantly suppressed when preceded by voltage steps to 0 mV or by evoked action potentials. The suppression was dependent on the presence of Ca2+ in the extracellular solution. Voltage steps to +100 mV were ineffective in suppressing NMDA responses. Therefore, the observed inactivation was caused by an increase in [Ca2+](i) following Ca2+ entry through NMDA channels or through voltage-gated Ca2+ channels. 4. Caffeine application reduced currents evoked by subsequent NMDA applications. This reduction was not dependent on the presence of extracellular Ca2+ but was abolished after incubation of the cells with ryanodine, suggesting that Ca2+ release from intracellular stores also induced CDI. 5. Simultaneous measurements of somal [Ca2+](i) and of currents evoked by somal NMDA applications showed that the magnitude of CDI was correlated with [Ca2+](i) levels and that [Ca2+](i) elevations of 100-300 nM were usually sufficient to inactivate NMDA currents by more than 30%. 6. Dose-response curves of non-inactivated and inactivated NMDA responses showed that the apparent receptor affinity for NMDA is not different under the two conditions. CDI is caused instead by non-competitive inhibition of NMDA receptors. CDI was not overcome by increasing glycine concentration, suggesting that it is not mediated by glycine dissociation from the receptor. 7. These results show that, with an intact intracellular environment, CDI in dorsal horn neurons constitutes a potent, inhibitory control of NMDA currents with a faster onset than previously demonstrated. CDI is induced by a variety of [Ca2+](i)-elevating stimuli of physiological relevance including Ca2+ entry through ligand- and voltage-gated channels and Ca2+ release from intracellular stores. Our demonstration that CDI is strongly expressed in neurons maturing in vivo supports the hypothesis that CDI may regulate, in part, the postsynaptic integration of excitatory input in the mature or maturing nervous system.