The discovery that blockade of N-methyl-D-aspartate (NMDA) receptors protects brain tissue against ischaemic damage has triggered enormous interest; and with the advance of intracerebral microdialysis, hundreds of studies have investigated changes in the extracellular levels of glutamate and other neurotransmitters during and after cerebral ischaemia. This work has made it apparent that the current concept of ischaemia-induced excitotoxicity, centred on excessive efflux of glutamate from nerve terminals, fails to correspond with reality since it conflicts with a number of key findings:(a) Excessive effluxes during ischaemia are not specific to excitatory amino acids--inhibitory transmitters are released to a similar extent;(b) neuronal death can occur several hours after a short ischaemic episode, whereas glutamate and aspartate accumulation in the neuronal microenvironment is cleared within minutes of reperfusion;(c) the penumbra is most receptive to cerebroprotection with glutamate receptor antagonists, but extracellular glutamate levels may not reach critical levels in this region; and (d) postischaemic treatment with glutamate receptor antagonists were neuroprotective in a number of studies. It has also become evident that most of the glutamate released in ischaemia is of metabolic origin, which questions the validity of therapeutic strategies aimed at preventing or reducing excessive release of neurotransmitter glutamate in ischaemia. However, the possibility that glutamate changes at the synaptic level may be small but pathologically important cannot be totally refuted. Apart from increased extracellular glutamate, the exceptional complexity of glutamate-operated ion channels can give rise to many potentially damaging mechanisms. Of particular interest are the possibilities of recurrent spreading depression in focal ischaemia, widespread and persistent strengthening of glutamatergic transmission, and abnormal modulation of the NMDA receptor-ionophore complex. There is also considerable evidence that, in certain brain regions, monoamines or their metabolic by-products may become neurotoxic either directly or from interplay with glutamatergic systems. All these processes deserve further examination to identify the most damaging and to indicate possible methods of intervention.