25 patients have been operated on by means of extra-intracranial anastomosis (22 with STA-MCA anastomosis, 3 with occipital-MCA anastomosis). 23 patients underwent an angiographic study early after surgery (two weeks). The patency rate is 14 out of 23 (13 STA, 1 occipital anastomosis). In comparison of its pre-operative size, the afferent artery has enlarged in the most cases, particularly in patients with complete obliteration, either of the carotid artery, or of the middle cerebral artery. In every case, only a limited part of the MCA territory is visualized through the anastomosis. In no case was the complete MCA field visualized; the frontal branches, particularly, are not supplied through a temporal anastomosis. In the case of occipital anastomosis, both upper and lower branches of MCA are supplied through the new channel. 11 patients underwent a second angiographic study, from one year through 28 months after the first one. In three patients with no patency on the first angiography the anastomosis remained non patent. So, in this series, no anastomosis was seen to became patent secondarily. In 8 patients with patency on the first control, the anastomosis remained patent on the second angiography. In patients with a pre-operative stenosis, no increasing of the size of the vessel could be noticed. The filling of the MCA branches is difficult to be discussed, for in these cases, the angiographies were not performed selectively through the external carotid artery. In patients with a pre-operative thrombosis, an enlarging of the vessels was seen, as well as an extension of the intra-cranial filling through the anastomosis. Clinical correlations are the following ones: the most patients with TIA's had a stenosis. They presented no increasing of the size of the vessels. They were doing well after operation, as if a little more of blood supply was sufficient to improve the general blood perfusion. Every patient with stroke had a pre-operative thrombosis and presented an enlarging of the vessels with a better filling on the second angiography, as if a great deal of additional blood supply was required; the clinical improvement is slow (3 out of 5) and remains often incomplete.