Thr authors report observations relative to four cases of traumatic carotid-cavernous fistulas for which angiographic pictures were similar. Autopsy examination permitted in two cases to establish that the tentorial branch of the meningo-hypophyseal trunk was involved in the genesis of both fistulas. One case was from a rupture, the other from tearing of that branch. Further examination of these two cases showed an absence of sphenoid bone fracture which suggested that any arterial perforation or shearing process could be excluded. Thus the pathogenic hypothesis of PARKINSON is further sustained. The authors suggest that the observed arterial lesions may be due to tensions occuring at the proximal part of branches of the meningo-hypophyseal trunk. Such tensions may be carried out through tractions on the posterior meningeal walls of the cavernous sinus when trauma occur on the posterior temporal area and the petrous bone. Such traction forces were shown to be real in one case where a fracture of the dorsum sellae and a bilateral caroti-cavernous fistula were caused by a bilateral temporo-petrous trauma.