The cerebral blood volume (CBV) has been measured by transverse axial tomography. The attenuation coefficient of 100 "pixels" located in a well determined region was measured before and after injection of a contrast material (Telebrix 38). The attenuation coefficient of blood samples was measured by CT in a phantom. Blood samples were drawn before and after injection at the same time as the tomographic section (time: half rotation of the scanner). The volume was calculated by the (formula: see text). The values obtained with very early measurements (1--5 min after injection), were very close to those measured by other methods (3--5 ml/100 ml). For 5 patients, the CBV was measured by the same method at different times: 20 measures between 2 and 50 min after the injection. The CBV seemed to vary in function with time according to the equation: CBV in ML%: ate-bt + Vo (t = time in minutes: a = integration constant, a = 1.94; b = time constant, b = 0.089; Vo = real CBV). The apparent variation of CBV was the result of two factors related to the blood-brain barrier, (1) water movement from the extravascular space towards the intravascular space, related to the hyperosmolarity of the contrast medium. The variations of osmolarity and haematocrit values measured after contrast injection did confirm this hypothesis; and (2) iodine movement from the intravascular space towards the extravascular space, due to the toxicity of the material (hyperosmolarity, Na ions, etc). The measurement of CBV by CT is only possible if the contrast materials do not modify the blood-brain barrier. The use of actual materials allows the study of the blood-brain barrier behaviour and enables to make an estimate of the cerebral extra-cellular space. But CT is not a right method of CBV measurement with the contrast medium in use nowadays.