Abstract The merit of using accurate measurements of the thalamostriate vein in patients with subarachnoid haemorrhage in order to depict ventricular size has been demonstrated. Ventricular dilatation following subarachnoid haemorrhage seems to be much more common than is generally realised. Nineteen out of 50 patients suffering from subarachnoid bleeding were found to have a significant ventricular dilatation as judged by measurements of the thalamostriate vein. Nine of the 19 had symptoms attributable to hydrocephalus; in 3 the symptoms subsided spontaneously. It is suggested that hydrocephalus develops in two stages following subarachnoid haemorrhage: first there is a reversible stage caused by partial obstruction of subarachnoid pathways. In some cases, however, hydrocephalus may become permanent because of the development of irreversible adhesions. The therapeutic measures evolving from this suggestion are discussed.