Etiology, mechanism and treatment remain controversial in aqueductal stenosis. The review of 114 cases operated on between 1975 and 1982 in the Service of Pediatric Neurosurgery of "Les Enfants-Malades" in Paris was undertaken with the hope of improving our understanding of these problems. Toxoplasmosis was in our series the most frequent etiology, accounting for 15% of the cases; 74% of the aqueductal stenoses in this series were of unknown origin. In two cases, a small arachnoid cyst, developed in contact with the ambient cistern, was the cause of the aqueductal stenosis. In about half of the 32 pneumoencephalograms performed, a rounded, dilated ambient cistern was found. No such dilatation was observed in 35 pneumoencephalograms performed in cases of communicating hydrocephalus. It is likely that some cases of aqueductal stenoses are the consequence of a compression of the brain stem by an overpressurized ambient cistern, whether communicating or not with the subarachnoid spaces. Although the ventricular volume is less reduced after percutaneous ventriculostomy than after shunting, the long term results of the two treatments are comparable. The risk of infection is lower with ventriculostomies, but the rate of failures is higher. Since ventriculocisternostomy is a safe procedure when patients are properly selected, it can be tested first, a shunt being inserted at a later time in case of failure.