Objectives:To describe the patterns of nosocomial infections in patients with traumatic injuries and to compare the associations between injury severity, derived from various severity scoring systems, and subsequent nosocomial infections.Design:Prospective observational study.Setting:A 750-bed university hospital serving as a medical school and referral center for the southern part of Thailand.Participants:All trauma patients admitted to the hospital for more than 3 days during 1996 to 1999 were eligible for this study.Methods:The severity of injuries was measured in terms of injury severity score (ISS), revised trauma score (RTS), new injury severity score (NISS), and trauma injury severity score (TRISS). Infections acquired during hospitalization were categorized using Centers for Disease Control and Prevention criteria. The association between severity of injury and nosocomial infection was examined with Poisson regression models.Results:There were 222 nosocomial infections identified among 146 patients, yielding an infection rate of 0.8 infections per 100 patient-days. Surgical-site infection was the most common site-specific infection, accounting for 31.1% of all infections. The incidence of intravenous catheter–related bloodstream infection was 1.6 infections per 100 catheter-days. The bladder catheter–related urinary tract infection rate was 2.8 infections per 100 catheter-days. The rate of ventilator-associated pneumonia was 3.2 infections per 100 ventilator-days. The incidence of infection correlated well with injury severity. The infection incidence rate ratios for one severity category increment of ISS, NISS, RTS, and TRISS were 1.65 (95% confidence interval [CI95, 1.42 to 1.92), 1.79 (CI95, 1.55 to 2.05), 1.64 (CI95, 1.43 to 1.88), and 1.32 (CI95, 1.14 to 1.52), respectively.Conclusions:Surgical-site infection was the most common site-specific nosocomial infection. The NISS might be the most appropriate severity scoring system for adjustment of infection rates in trauma patients.