This study examined the involvement of amygdala N-methyl-D-aspartate (NMDA) receptors in long-term retention of an inhibitory avoidance response. Rats bearing chronic cannulae implanted into the basolateral amygdala were trained on a one-trial inhibitory avoidance task and tested for retention 21 days later. They received intra-amygdala injections of vehicle (Veh) or 0.25, 1.25 or 5.0 micrograms of a competitive NMDA antagonist-2-amino-5-phosphonopentoic acid (AP5) either 5 min before training, immediately after training or 5 min before testing. Results indicated that pretraining intra-amygdala injections of AP5 at all doses impaired retention performance profoundly. Intra-amygdala injections of AP5 immediately after training caused a dose-dependent retention deficit: 0.25 microgram induced no deficit and 5.0 micrograms induced a great deficit. Immediate posttraining intra-amygdala injections of a non-competitive NMDA antagonist MK-801, also impaired retention but MK-801 given 2 hrs after training had no significant effect. In contrast to the marked amnestic effect produced by pre- or posttraining intra-amygdala injections of AP5, intra-amygdala injections of AP5 given just before retention tests had no discernible effect on retention performance. The retention deficit induced by pretraining intra-amygdala injections of 1.25 micrograms AP5 was ameliorated completely by N-methyl-DL-aspartate (0.25 microgram), but also partially by norepinephrine (0.2 microgram) infused into the amygdala immediately after training. However, posttraining infusion 0.2 microgram norepinephrine failed to attenuate significantly the amnestic effect induced by 5.0 micrograms AP5. These findings, taken together, suggest that NMDA receptors in the amygdala are normally involved in memory formation processing of affective experience.