The left middle cerebral artery (MCA) was occluded at its origin via a subtemporal approach under operating microscopic control in 24 dogs. In 8 of these 24 dogs, end-to-side anastomosis between the maxillary artery (MA) and a branch of the left middle cerebral artery (MA•MCA anastomosis) was completed 4 hours after MCA occlusion (prompt bypass). In 5 dogs, MA•MCA anastomosis was performed 3 weeks after MCA occlusion (delayed bypass). These MA•MCA anastomoses were carried out with microsurgical technique as well. Remaining 11 dogs without MA•MCA anastomosis were used for control animals. Each animal was observed clinically every day until sacrifice. In control animals, common carotid angiography was performed just prior to sacrifice between 2nd and 5th postoperative weeks. Treated animals were studied with selective external carotid angiography 2 weeks after MA•MCA anastomosis. After sacrifice, transcarotid perfusion with 10% formalin solution was carried out, the brains were removed carefully, and they were additionally fixed in 10% formalin for 2 weeks. Each brain was sectioned, stained and examined histologically. 1. MA•MCA anastomosis was devised as a new experimental model for extracranial-intracranial bypass graft in the dog. In prompt bypass of 8 dogs, 7 cases showed patency of anastomosis (88%), and in delayed bypass of 5 dogs, angiogram revealed 4 patent anastomosis (80%). In successful cases, angiogram demonstrated excellent filling of the entire territory of MCA through the new shunts. Anatomical and spatial advantage of the MA seemed to have brought good results in our series of anastomosis than that of other previous experimental extracranial-intracranial bypass graft in the dog. This is the first report on successful patent bypass graft after long-term occlusion of the MCA in animals. It seems that this new experimental model brings the progress in the field of studies of extracranial-intracranial bypass graft for cerebral infarction. 2. Clinical evaluation of all control animals showed mild to severe neurological deficits, mainly contralateral hemiparesis, or death. On the other hand, animals with patent prompt bypass demonstrated no neurological deficits. Gross and histological evaluation of the brains showed that permanent occlusion of the MCA produced medium or large-sized infarct located in its territory in control animals, and patent prompt bypass usualy caused no or only microscopical infarct. In patent delayed bypass, the size of infarct seems to be smaller than that of non-treated animals. No hemorrhagic infarct caused by revascularization was found in treated animals with both prompt and delayed bypass grafts. There was fairly good relation between the neurological deficits and the pathological features in the brains. In general, it seemed that animals with patent bypass grafts fared better than untreated animals both clinically and pathologically. This experimental data suggest that reestablishment of blood flow by extracranial-intracranial bypass graft, especially in prompt bypass, might lead to significant restoration of neurological function without pathological damage of the brain.