In order to evaluate the contribution of computerized tomography (CT) to the diagnosis of cerebral thrombophlebitis, a series of 28 cases was reviewed and compared with data from the literature. In an examination carried out 4 to 5 days of its constitution the thrombus may be directly visualized as a spontaneous hyperdensity. This early but very transient sign, called "cord sign", can easily be overlooked, which explains why it was found in only 5 of our 28 cases and in 2% of the largest series of the literature. The thrombus thereafter becomes hypodense and can be intensified by peripheral contrast enhancement which produces the classical "delta sign". This sign is more frequent: 13/28 in our series and 16 to 30% in published cases. It is usually found in the superior sagittal sinus and must be distinguished from anatomical variations which are common at that level. These two direct signs acquire a greater value when associated with such indirect signs as diffuse or localized cerebral oedema (12 to 52%) and venous ischaemia (22 to 59%). Venous ischaemia is characterized by its strong bleeding potential (more than 50% of the cases) and by its usually favourable course; these two elements and its site differentiate it from arterial ischaemia. Finally, venous stasis is responsible, in 5 to 19% of the cases, for intense enhancement of the tentorium cerebelli; this sign is not specific but easy to evidence and of great value when associated with a direct sign. Dilatation of cortical veins, found in 4 of our 28 cases, also seems to be an interesting sign which, to our knowledge, has not yet been mentioned in the literature. Since in 3.6 to 26% of the cerebral thrombophlebitis the CT scan is normal, a negative CT examination does not rule out this disease, and in many cases the exploration must be rapidly completed by angiography or MRI. Because it is non-invasive and very sensitive to flows, MRI has become the key examination to assert the diagnosis. Angiography is now restricted to those cases where cases where MRI cannot be performed promptly or to certain, purely cortical thrombophlebitis which might pass unnoticed at MRI. When carried out and interpreted cautiously, angiography always shows the venous thrombosis, its exact size and its suppletive network. The results of this study show that MRI alone can diagnose cerebral thrombophlebitis in most patients, that CT well done and interpreted often provides useful but seldom sufficient indices, and that angiography should be reserved for difficult cases.