OBJECTIVE CSF shunt infection requires both surgical and antibiotic treatment. Surgical treatment includes either total shunt removal with external ventricular drain (EVD) placement followed by new shunt insertion, or distal shunt externalization followed by new shunt insertion once the CSF is sterile. Antibiotic treatment includes the administration of intravenous antibiotics. The Hydrocephalus Clinical Research Network (HCRN) registry provides a unique opportunity to understand reinfection following treatment for CSF shunt infection. This study examines the association of surgical and antibiotic decisions in the treatment of first CSF shunt infection with reinfection. METHODS A prospective cohort study of children undergoing treatment for first CSF infection at 7 HCRN hospitals from April 2008 to December 2012 was performed. The HCRN consensus definition was used to define CSF shunt infection and reinfection. The key surgical predictor variable was surgical approach to treatment for CSF shunt infection, and the key antibiotic treatment predictor variable was intravenous antibiotic selection and duration. Cox proportional hazards models were constructed to address the time-varying nature of the characteristics associated with shunt surgeries. RESULTS Of 233 children in the HCRN registry with an initial CSF shunt infection during the study period, 38 patients (16%) developed reinfection over a median time of 44 days (interquartile range [IQR] 19-437). The majority of initial CSF shunt infections were treated with total shunt removal and EVD placement (175 patients; 75%). The median time between infection surgeries was 15 days (IQR 10-22). For the subset of 172 infections diagnosed by CSF culture, the mean ± SD duration of antibiotic treatment was 18.7 ± 12.8 days. In all Cox proportional hazards models, neither surgical approach to infection treatment nor overall intravenous antibiotic duration was independently associated with reinfection. The only treatment decision independently associated with decreased infection risk was the use of rifampin. While this finding did not achieve statistical significance, in all 5 Cox proportional hazards models both surgical approach (other than total shunt removal at initial CSF shunt infection) and nonventriculoperitoneal shunt location were consistently associated with a higher hazard of reinfection, while the use of ultrasound was consistently associated with a lower hazard of reinfection. CONCLUSIONS Neither surgical approach to treatment nor antibiotic duration was associated with reinfection risk. While these findings did not achieve statistical significance, surgical approach other than total removal at initial CSF shunt infection was consistently associated with a higher hazard of reinfection in this study and suggests the feasibility of controlling and standardizing the surgical approach (shunt removal with EVD placement). Considerably more variation and equipoise exists in the duration and selection of intravenous antibiotic treatment. Further consideration should be given to the use of rifampin in the treatment of CSF shunt infection. High-quality studies of the optimal duration of antibiotic treatment are critical to the creation of evidence-based guidelines for CSF shunt infection treatment.