BACKGROUND: Ventilator-associated pneumonia (VAP) rates reported by the National Healthcare Safety Network (NHSN) are used as a benchmark and quality measure, yet different rates are reported from many trauma centers. This multi-institutional study was undertaken to elucidate VAP rates at major trauma centers. METHODS: VAP rate/1,000 ventilator days, diagnostic methods, institutional, and aggregate patient data were collected retrospectively from a convenience sample of trauma centers for 2008 and 2009 and analyzed with descriptive statistics. RESULTS: At 47 participating Level I and II centers, the pooled mean VAP rate was 17.2 versus 8.1 for NHSN (2006-2008). Hospitals' rates were highly variable (range, 1.8-57.6), with 72.3% being above NHSN's mean. Rates differed based on who determined the rate (trauma service, 27.5; infection control or quality or epidemiology, 11.9; or collaborative effort, 19.9) and the frequency with which VAP was excluded based on aspiration or diagnosis before hospital day 5. In 2008 and 2009, blunt trauma patients had higher VAP rates (17.3 and 17.6, respectively) than penetrating patients (11.0 and 10.9, respectively). More centers used a clinical diagnostic strategy (57%) than a bacteriologic strategy (43%). Patients with VAP had a mean Injury Severity Score of 28.7, mean Intensive Care Unit length of stay of 20.8 days, and a 12.2% mortality rate. 50.5% of VAP patients had a traumatic brain injury. CONCLUSIONS: VAP rates at major trauma centers are markedly higher than those reported by NHSN and vary significantly among centers. Available data are insufficient to set benchmarks, because it is questionable whether any one data set is truly representative of most trauma centers. Application of a single benchmark to all centers may be inappropriate, and reliable diagnostic and reporting standards are needed. Prospective analysis of a larger data set is warranted, with attention to injury severity, risk factors specific to trauma patients, diagnostic method used, VAP definitions and exclusions, and reporting guidelines. LEVEL OF EVIDENCE: III, prognostic study.