The efficacy of GM1 ganglioside treatment in stroke was studied in a permanent middle cerebral artery occlusion model in the rat. A dose-dependent attenuation of infarct size at 24 hr was documented with the maximum effective dose halving the volume of the experimental stroke. Delayed administration at 5 min, but not 15 min, after vessel occlusion was as effective as preocclusion drug administration. Morphologic sparing was confined to the cortical penumbra; no protection in the ischemic core was found. Morphologically salvaged cortex was also metabolically preserved as demonstrated by quantitative measurement of glucose utilization. In vivo microdialysis demonstrated an attenuation of ischemic-induced glutamate release in the cortex with GM1 administration but no effect was found in the caudate. Hypotension did not occur even with doubling of the maximally effective dose of GM1. Accordingly, GM1 may be a safe and effective treatment for stroke.