Cultured cerebellar granule cells and cerebellar slices from neonatal rats have been widely used to examine the biochemistry of excitatory amino acid-induced cell death mediated in part by the activation of NMDA receptors. However, the NMDA subunit stoichiometry, producing functional NMDA receptors is different in cultured granule cells, neonatal and adult rat cerebellum as compared to the NMDA receptors in forebrain regions. We have used the L-2-chloropropionic acid (L-CPA) (750 mg/kg) model of NMDA-mediated selective cerebellar granule cell necrosis in vivo to examine the role of the glycine binding site and possible effect of the NR2C subunit (which is largely expressed only in the cerebellum) on granule cell necrosis. The abilities of various NMDA receptor antagonists were examined in vivo to determine the relative contribution of both glutamate and glycine sites involved in the L-CPA-induced neurotoxicity. The potent neuroprotective, non-competitive NMDA receptor antagonist dizocilpine (MK-801) was compared with glutamate and glycine site NMDA antagonists. We have examined a number of markers for the L-CPA-induced granule cell necrosis. The L-CPA-induced reduction in cerebellar aspartate and glutamate concentrations were used as markers of granule cell necrosis. We also measured the cerebellar water content and sodium concentrations as measures of the L-CPA-induced cerebellar edema that accompanies the granule cell necrosis. Finally the ability of the NMDA antagonists to attenuate the L-CPA-induced reductions in body weight gain and the prevention of the loss in hindlimb function using a behavioral measure of hindlimb retraction were examined. The potent glutamate antagonists, CPP and CGP40116 and dizocilpine prevented the L-CPA-induced locomotor dysfunction and granule cell necrosis as measured by their ability to prevent L-CPA-induced reduction in aspartate and glutamate concentrations. CPP, CGP40116 and dizocilpine also prevented the appearance of cerebellar edema following L-CPA administration. In addition, dizocilpine, CPP and CGP40116 were able to partially prevent the L-CPA-induced loss in body weight over the 48 h experimental period. In contrast, none of the glycine partial agonists or antagonists, namely (+/-)HA-966, D-cycloserine, MDL-29,951, DPCQ, MNQX or L-701 252 were able to prevent the L-CPA-induced loss in body weight, L-CPA-induced granule cell necrosis and behavioral disturbances when administered to rats. None of the NMDA antagonists had any effect on the cerebellar neurochemistry when injected alone or had any effect on animal behavior except for dizocilpine, CPP, CGP40116 and (+/-)HA-966 which resulted in a transient sedation for between three and five hours immediately following their administration. In conclusion, we demonstrated that NMDA open channel blockade and glutamate antagonists can provide full neuroprotection against the L-CPA-induced granule cell necrosis. The failure of the glycine partial agonist and antagonists to provide any neuroprotection against L-CPA-induced neurotoxicity in the cerebellum contrast with their neuroprotective efficacy in other animal models of excitatory amino acid-induced cell death in forebrain regions in vivo. We therefore suggest that the glycine site plays a lesser role in modulating NMDA receptor function in the cerebellum and may explain why cells expressing NMDA receptors composed of NR1/NR2C subunits are particularly resistant to excitatory amino acid-induced neurotoxicity.