Background. Neurologic injury associated with prolonged hypothermic circulatory arrest (HCA) may be mediated by calcium-dependent glutamate excitotoxicity (GE). The monosialoganglioside GM1 has been shown in vitro to limit GE in conditions of metabolic stress. To test the hypothesis that gangliosides can prevent HCA-induced brain injury, GM1 was used in a canine model of HCA. Methods. Twelve male dogs were placed on closed-chest cardiopulmonary bypass, subjected to 2 hours of HCA at 18° C, and rewarmed to 36° to 37° C on closed-chest cardiopulmonary bypass. All were mechanically ventilated and monitored for 20 hours before extubation and survived for 3 days. Group 1 dogs (n = 6) were pretreated with GM1, 30 mg/kg/24hr for 3 days before HCA, and received continuous infusion of GM1 during the procedure and 30 mg/kg/24hr for 3 days after HCA. Group 2 dogs (n = 6) received vehicle only. With a species-specific behavior scale that yielded a neurodeficit score ranging from 0% (normal) to 100% (brain dead), all animals were neurologically assessed every 12 hours. After death at 72 hours, brains were examined by glutamate receptor autoradiography and by histologic examination for patterns of selective neuronal necrosis and were scored blindly from 0 (normal) to 100 (severe injury). Results. Group 1 dogs had better neurologic function compared with group 2 (neurodeficit score, 4.2% ± 3% vs 38.4% ± 8%; p < 0.001) and had less neuronal injury (11.3 ± 3 vs 48.3 ± 9, p < 0.001). Densitometric receptor autoradiography revealed preservation of neuronal glutamate receptor expression in group 1 only. Conclusions. These results provide evidence of a role for GE in the development of HCA-induced brain injury and suggest that monosialogangliosides may have a neuroprotective capacity in prolonged periods of HCA.