1. (1) Since 1960, 104 circles of Willis have been studied macroscopically. All the patients, operated upon or not, died after rupture of berry aneurysms. 126 of these aneurysms have been demonstrated. 46 were on the anterior communicating artery (36.8%), 42 on the internal carotid (33.6%), and 30 on the middle cerebral artery (24%). More of the patients were female than male and most were between 40 to 70 years old. Multiple saccular aneurysms were encountered in 15% of cases. 2. (2) Anterior communicating aneurysms: 31 out of 46 arose from the anterior communicating artery, and 14 from both this vessel and a pericallosal artery. There is no relationship between the presence of such an aneurysm and the size of anterior cerebral arteries. Nevertheless in 26 cases, there was a difference between the two sides (4 filiform arteries and 3 cases with only one anterior cerebral artery). Heubner's artery begins nearly always at the level of anterior communicating artery in cases with saccular aneurysm. If the anterior communicating artery is double, the malformation is usually inserted on the superior branch, whereas, if a median artery supplying the corpus callosum is demonstrated, it starts, on the contrary, from the inferior branch. Only 7 cases had a false sac. Major anomalies of the circle of Willis were found in 20% of cases. 3. (3) Among the 42 internal carotid saccular aneurysms, 38 were of the supraclinoid type, 4 were located at the bifurcation, 10 arose from both the internal carotid and posterior communicating arteries. Here too, no ratio could be demonstrated between the size of the posterior communicating artery and the presence of such a saccular aneurysm. The average width (10 mm) of the malformation is larger than elsewere, but the size of the neck (3 mm) is the same everywhere. In 40% of cases, a false sac has been encountered. No internal carotid thrombosis was found, either natural or post-surgical. 4. (4) Middle cerebral aneurysms: 28 out of 30 were inserted at the bifurcation and 17 had no connection with secondary branches. In 20% of cases, a small collateral vessel was more or less adherent to the sac. In two-thirds of cases, there was no false sac. 5. (5) Whatever the location of the aneurysm: (a) no collateral branch from the sac has been demonstrated. In a few cases, a small artery arising from the main trunk, was more or less adherent to the sac, in part of its course; (b) All the ruptures were apical; (c) Some patients had been operated upon and most of the deaths were unexplained macroscopically (as by a clip badly located, thrombosed artery, intracerebral clot or softening); (d) Localized or diffuse atheroma was noticed in 40% of cases. Both were found in only 26%. 6. (6) Certain therapeutic deductions can be attempted, knowing the location of ruptures, the lack of collateral branches and details of the origin of saccular aneurysms.