OBJECTIVE To evaluate the agreement between the current National Healthcare Safety Network (NHSN) definition for ventriculitis and others found in the literature among patients with an external ventricular drain (EVD) DESIGN Retrospective cohort study from January 2009 to December 2014 SETTING Neurology and neurosurgery intensive care unit of a large tertiary-care center PATIENTS Patients with an EVD were included. Patients with an infection prior to EVD placement or a permanent ventricular shunt were excluded. METHODS We reviewed the charts of patients with positive cerebrospinal fluid (CSF) cultures and/or abnormal CSF results while they had an EVD in place and applied various ventriculitis definitions. RESULTS We identified 48 patients with a total of 52 cases of ventriculitis (41 CSF culture-positive cases and 11 cases based on abnormal CSF test results) using the NHSN definition. The most common organisms causing ventriculitis were gram-positive commensals (79.2%); however, 45% showed growth of only 1 colony on 1 piece of media. Approximately 60% of the ventriculitis cases by the NHSN definition met the Honda criteria, approximately 56% met the Gozal criteria, and 23% met Citerio's definition. Cases defined using Honda versus Gozal definitions had a moderate agreement (κ=0.528; P<.05) whereas comparisons of Honda versus Citerio definitions (κ=0.338; P<.05) and Citerio versus Gozal definitions (κ=0.384; P<.05) had only fair agreements. CONCLUSIONS The agreement between published ventriculostomy-associated infection (VAI) definitions in this cohort was moderate to fair. A VAI surveillance definition that better defines contaminants is needed for more homogenous application of surveillance definitions between institutions and better comparison of rates.