Abstract
In two cases of cerebrospinal fluid rhinorrhoea in which scinticisternography failed to identify the fistulae, the tracts were demonstrated by positive contrast ventriculography. It is postulated that the fistula communicated with the ventricles but was isolated from the subarachnoid space by adhesions (demonstrated at operation in one case). There was `high pressure rhinorrhoea' in one case. The rhinorrhoea ceased after insertion of ventriculoatrial shunt.