A wealth of experimental evidence demonstrates that cerebral ischemia causes excessive release of glutamate and that glutamate contributes to ischemic injury. Glutamate antagonism by any of several mechanisms can ameliorate the extent of infarction. These antagonists comprise noncompetitive blockers of the ion channel associated with the N-methyl-D-aspartate (NMDA) receptor [e.g., aptiganel (Cerestat)], competitive antagonists of the glutamate recognition site of the NMDA receptor (e.g., selfotel) or of the glycine recognition site (e.g., ACEA 1021, GV150526), antagonists at the polyamine site (e.g., eliprodil), and drugs that may interfere with glutamate release by sodium channel blockade as well as having other actions (e.g., lubeluzole, 619C89). Clinical experience suggests that although some NMDA antagonists are poorly tolerated at putative neuroprotective doses (e.g., selfotel), potentially neuroprotective plasma concentrations can be achieved in humans with others (e.g., aptiganel), though tolerable adverse effects are frequently observed. These clinical effects include hypertension (which is probably preferable to the hypotension seen with nimodipine and lifarizine), sedation, confusion or hallucinations and, at high doses, catatonia. Glycine antagonists may be associated with fewer adverse effects, but preclinical studies suggest that brain penetration may be low. Although recent studies with selfotel and eliprodil have been discontinued because of insufficient evidence for a satisfactory risk/benefit ratio, encouraging experience with aptiganel, magnesium, and glycine antagonists has prompted continued clinical trials with these agents. To be of sufficient size to detect a clinically useful improvement in outcome, these trials need to be large (600-1,000 patients). Present trials with aptiganel (Cerestat) are comparing the efficacy and tolerability of two doses vs. placebo in patients treated within 6 hours of ischemic stroke. Outcome is assessed by the modified Rankin Scale at 3 months.