The non-competitive N-methyl-D-aspartate (NMDA) antagonist MK-801, injected intraperitonealy at 0.1 mg/kg, at times between 1 h before and 5 min after training chicks on a one-trial passive avoidance task, resulted in amnesia for the task on test 3 or 24 h subsequently. No amnesia was apparent at 24 h if chicks were injected between 1 and 6 h after training. Amnesia did not develop immediately; it was not apparent 30 min after training in chicks injected 5 min after training. At this dose of MK-801 no other effects on motor or pecking behaviour of the birds were observed. Bilateral or unilateral intracerebral injections of 1.5 nM MK-801 5 min after training produced a similar amnesia at 3 h to that of intraperitonealy injected MK-801; no hemispheric differences were observed, presumably because of the ready diffusion of the MK-801. By contrast, intracerebral injections of the non-NMDA glutamate antagonists 6-cyano-7-nitroquinoxaline-2, 3-dione (CNQX), 6,7-dinitroquinoxaline-2,3-dione (DNQX) and 6,7-nitro-7-sulphamoyl-benzoquinoxaline-2,3-dione (NBQX) (0.066 μM) 5 min after training, despite producing severe if transient behavioural disturbances, were without effect on retention for the avoidance response in chicks tested 3 h subsequently. We interpret these results as pointing to a requirement for NMDA, but not kainate or quisqualate, receptor activation as an early enabling event in the biochemical cascade required for long-term memory formation for passive avoidance in the chick.