A retrospective study of the management of patients with infected cerebrospinal fluid (CSF) shunts was undertaken, covering the 20 years from 1960 to 1979, inclusive, and involving 222 patients with 267 infections. The data were analyzed with emphasis on influences surrounding treatment choice and subsequent outcome. Treatment was classified into three major categories: medical management (antibiotics alone), surgical management (antibiotics plus operative removal of the infected shunt), and no treatment (ranging from admission and observation only to shunt revision), the diagnosis of shunt infection having been missed. Results showed surgical treatment to be more efficacious than medical or no treatment, with a higher rate of initial cure, and lower morbidity and mortality rates. Also examined were the relationships among clinical presentation, infection rate, and results of specimens sent for culture, and initial treatment. The definitive nature of initial treatment was revealed to be directly proportional to the aggressiveness of microbiological investigation. This latter aspect was related to clinical presentation, with shunt malfunction being the least recognized symptom of shunt infection. Patients presenting with blocked shunts were less likely to receive therapy appropriate for infection than any other group, leading to the conclusion that shunt malfunction may be more specific to infection than heretofore believed.