Recent data suggest that brain damage in ischemia, hypoglycemia, and several other brain diseases is caused by excitotoxic mechanisms which are triggered by presynaptic release of glutamate and related excitatory amino acids, and which involve an abnormal postsynaptic influx of calcium into cells containing a high density of glutamate receptors. This contention is supported by results demonstrating reduction of infarct size in focal ischemia due to middle cerebral artery (MCA) occlusion, and amelioration of neuronal necrosis in hypoglycemic coma, by antagonist which block the NMDA type of glutamate receptor. These results underscore the pathogenetic role of calcium influx into energy-compromised cells since the NMDA receptor-linked ion channel has a high conductance to calcium. The issue has been clouded by the inability of NMDA antagonists to ameliorate brain damage due to cardiac arrest, or to forebrain ischemia in rats and gerbils. In these conditions, however, an AMPA receptor blocker (NBQX) has been found efficacious. These results demonstrate that the pathophysiology of ischemic lesions is different in the cardiac arrest type of ischemia and in lesions due to MCA occlusion, and demand an explanation of the differences in therapeutic response. Tentatively, the cardiac arrest type of ischemia is so dense that multiple calcium conductances are activated in the energy-deprived tissue, explaining why any drug which acts on only one of them (such as an NMDA antagonist) cannot prevent cellular calcium overload. Furthermore the ultimate brain damage, which is often conspicuously delayed, may be secondary to upregulation of synaptic efficacy, causing increased calcium cycling and calcium-related damage. In this situation, an AMPA receptor blocker may be efficacious because it blocks "fast" excitation and Na+ influx, an "upstream" event which causes "downstream" calcium influx via multiple pathways. In the perifocal ("penumbra") zone of a stroke lesion, the situation is different since depolarisation is initially moderate and/or intermittent. Furthermore, since ATP is still produced (albeit at a reduced rate) the problem is one of a disturbed pump/leak relationship. Then, blockade of a major calcium-carrying channel by NMDA receptor blockers, or of the trigger to depolarisation by an AMPA receptor antagonist, may improve the pump/leak relationship and carry cells in the penumbra over a critical period.