Twelve patients between the years of 1964 and 1971 underwent surgical treatment for the so-called Dandy-Walker syndrome at the Johns Hopkins Hospital. Five patients belonged to an early group (1-8 weeks); four patients to an intermediary group (3-20 months); and a late group consisted of three patients (3 years, 14 years and 34 years). The value of the various diagnostic studies is discussed on the basis of an analysis of the plain skull X-rays, cerebral angiograms, and combined air studies. The importance of a careful delineation of the underlying pathological condition is stressed to enable the surgeon to choose the appropriate technique. The diagnostic studies confirmed the diagnosis of a true Dandy-Walker syndrome in six of the twelve cases. The remaining six proved to have either an independent posterior fossa cyst with communication of the subarachnoid space and occasionally with the ventricles; or an isolated posterior fossa cyst with no communication but with impingement on the aqueduct and on the midline structures. In six patients, a primary simple shunt procedure was performed which had to be followed by a posterior fossa craniectomy in one and by a double-shunt (ventricle and posterior fossa cyst) in another. Four patients underwent a posterior fossa craniectomy; as the primary and only procedure (2), craniectomy followed by a ventriculoperitoneal shunt (1), or an unsuccessful shunt followed by craniectomy (1). In two patients, a so-called 'double shunt' was performed as a second operation, draining simultaneously the right lateral ventricle and the posterior fossa cyst. There was no death in this series. Three children had evidence of infected shunts; they responded to adequate therapy. Four had excellent results, with normal neurological examination and normal development. Four had good results, performing in school (working) was markedly improved; two had fair results with residual neurological deficit and some mental retardation. Two patients had to be considered as poor results. Both of them, however, have shown evidence of concomitant severe congenital abnormalities of the cerebellum on the diagnostic studies. The follow-up period ranged from 1 1/2 to 7 1/2 years.