We concluded a prospective epidemiologic study of ventriculostomy-related infections (ventriculitis or meningitis) in 172 consecutive neurosurgical patients over a two-year period to determine the incidence, risk factors, and clinical characteristics of the infections. Ventriculitis or meningitis developed in 19 of 172 patients (11 per cent) undergoing a total of 213 ventriculostomies. When data from all these cases plus five cases of nonventriculostomy-related infection were combined, cerebrospinal-fluid pleocytosis was more significantly associated with the diagnosis of ventriculitis or meningitis (P less than 0.0001) than were fever and leukocytosis (P = 0.07). Risk factors for ventriculostomy-related infections included intracerebral hemorrhage with intraventricular hemorrhage (P = 0.027), neurosurgical operations (P = 0.016), intracranial pressure of 20 mm Hg or more (P = 0.019), ventricular catheterization for more than five days (P = 0.017), and irrigation of the system (P = 0.021). Previous ventriculostomy did not increase the risk of infection with subsequent procedures. We conclude that ventriculostomy-related infections may be prevented by maintenance of a closed drainage system and by early removal of the ventricular catheter. If monitoring is required for more than five days, the catheter should be removed and inserted at a different site.