Increased glutamate-receptor-mediated Ca++ influx is considered an important factor underlying delayed neurodegeneration following ischemia or seizures. Until recently, the NMDA receptor was the only glutamate receptor known to be Ca(++)-permeable. It is now well established that glutamate receptors of the AMPA type, encoded by a gene family designated GluR1-GluR4, exist in both Ca(++)-permeable and Ca(++)-impermeable forms, depending on their subunit composition and degree of RNA editing. Recombinant channels assembled without GluR2 are permeable to Ca++; channels assembled with (edited) GluR2 are Ca(++)-impermeable. AMPA receptors in most adult neurons are hetero-oligomers containing GluR2 subunits, but some neurons have GluR2-less, Ca(++)-permeable receptors. The "GluR2 hypothesis" predicts that a relative reduction in the expression of GluR2 results in enhanced Ca++ influx through newly synthesized AMPA receptors, thereby increasing neurotoxicity of endogenous glutamate. Recent observations indicate reduction in GluR2 expression and predict formation of Ca(++)-permeable AMPA receptors following global ischemia and kainate-induced status epilepticus; these changes are likely to be a major factor contributing to the delayed neurodegeneration that follows these pathological events. The delayed neurodegeneration appears to be primarily apoptotic. Thus, there are at least three strategies for neuroprotection: block of formation of GluR2-less receptors, which may be possible at several levels; block of the GluR2-less receptors themselves; and block of the subsequent apoptosis.