The clinical observation of 47 patients, undergoing surgery for "normotensive hydrocephalus" during the period January 1975 to January 1981 have been analysed retrospectively. The condition of half the patients improved following insertion of a cerebrospinal fluid bypass. Complications of varying degrees of severity were noted in 21 percent of cases. The correlation between the post-operative results and the clinical data was subjected to statistical analysis. Some factors appeared to have an important predictive value. These were the age, the absence of cortical atrophy on C.T. scan and the prolonged ventricular filling on cisternography. Other factors, although without statistical significance, seem to predict a better result: the existence of prior etiological data, presence of a frontal syndrome and evidence of periventricular hypodensity on C.T. scan. Finally, some other factors in our series appeared to have no influence on the prognosis. These were: duration of the preoperative evolution of the disorder, the degree of mental disturbance and difficulties of gait, the presence of urinary incontinence, the value of the ventricular cranial ratio. None of the recognized significant criteria was sufficiently reliable to predict the quality of surgical result, when studied in isolation. The association between the different significant criteria, such as have been described above notably, improves the chances of a favorable post-operative outcome but it is insufficient to predict it with great certitude. The continuous monitoring of the intracranial pressure during periods of 24 to 48 hours appears to be the most reliable factor on which to make a decision to operate in the light of the findings described in recent literature.