The incidence of visual failure with carotid-cavernous fistula is high. It was impaired in 89 per cent. of the cases studied by de Schweinitz and Holloway (I908) and in 73 per cent. of those studied by Sattler (I920). The latter recorded blindness or near blindness in almost half of his patients. Treatment has reflected a clinical preoccupation with elimination of the bruit and reduction of the proptosis, and modern methods have become increasingly proficient at obtaining this by more extensive surgery (Dandy, I935; Echols and Jackson, I959; Hamby, I966). However, in this condition, where the threat to life is small, preservation of vision becomes the major aim of therapy. Realisation of this goal has been lacking, though the ocular and cerebral hazards of surgery have recently been stressed by neurosurgeons (Walker and Allegre, I956; Stern, Brown, and Alksne, I967). Many articles and monographs have described the ocular complications of carotidcavernous fistula both before and after surgery, but the patho-physiology of these changes has never received the detailed and systematic attention it deserves. We have approached this problem by analysing critically the preoperative and postoperative causes of impaired vision in a series of 25 carotid-cavernous fistulae studied and treated at the University of California Medical Center during the past Io years. The changes that occurred in the eye, from the cornea to the optic nerve, were assessed in regard to: (i) Their effect on visual function. (2) Their appearance and resemblance to the hypoxic complications seen in other vascular diseases. (3) Their improvement or deterioration after neurosurgical procedures which alter the circulatory dynamics of the eye and orbit. Abnormalities in ocular perfusion, circulation time, and vascular permeability were recorded in selected cases by ophthalmodynamometry or fluorescein angiography.