Rats were bilaterally implanted with cannulae in the entorhinal cortex, amygdala, and hippocampus; after recovery, they were trained in a step-down inhibitory avoidance task and tested for retention 24 h later. Muscimol (0.03 microgram) or D-amino-5-phosphonovalerate (5.0 micrograms) infused in the entorhinal cortex 20 min prior to training inhibited the amnestic effect of the same dose of muscimol infused into this area 100 min after training. Thus, memory-relevant information must be processed by the entorhinal cortex at the time of training in order that this cortex may play a late post-training role in memory processing. Pretraining intraentorhinal muscimol administration did not affect the amnestic effect of the post-training infusion of muscimol into the amygdala and hippocampus, or the inhibition of memory expression induced by a pretest infusion of CNQX into the amygdala and hippocampus or into the entorhinal cortex. Pretest intraentorhinal muscimol also did not influence the effect of pretest intra-amygdala and intrahippocampal CNQX administration. These data indicate that the cells of the entorhinal cortex that are sensitive to pretraining muscimol are not part of the inputs that lead to post-training processing by the amygdala and hippocampus, or to the intervention of the amygdala, hippocampus, and entorhinal cortex in memory expression. The present findings are compatible with the possibility that, instead, the entorhinal cortex may be an output of the amygdala and hippocampus at the time of memory expression.