Intrastriatal injection of the glutamate analogue N-methyl-D-aspartate (NMDA, 25 nmol) in postnatal day (PND) 7 rats provides a rapid, sensitive, and reproducible assay in which potential neuroprotective strategies against excitotoxic neuronal injury can be examined in vivo. Brain injury is quantified 5 days postinjection by comparison of the weights of the injected and contralateral cerebral hemispheres. Intraperitoneal injections (15 minutes post-NMDA) of competitive and noncompetitive NMDA receptor antagonists attenuated the severity of NMDA-induced brain injury. The rank order of neuroprotective potency of these antagonists was CGS-19755 greater than DOIPG greater than dextromethorphan greater than HA-966. Of these compounds only the competitive antagonist CGS-19755 provided complete neuroprotection. NMDA-mediated brain injury was also reduced by the specific sigma receptor ligands +PPP and haloperidol (35% reduction). In contrast, drugs that reduce presynaptic neurotransmitter release (adenosine) or enhance neuronal inhibition (baclofen) were not effective against NMDA toxicity. Although all five of the anticonvulsants tested limited NMDA-induced seizure activity, only carbamazepine reduced NMDA-mediated brain injury (36% reduction). These findings extend earlier observations that NMDA receptor antagonists can limit NMDA-induced toxicity in vivo and suggest that sigma receptors contribute to the pathophysiology of NMDA-mediated brain injury in vivo. Furthermore, NMDA-induced seizures and brain injury appear dissociable in this in vivo model. The results illustrate important practical limitations of neuroprotection in vivo vs. in vitro.