Vancomycin-resistant enterococcus (VRE) and methicillin-resistant Staphylococcus aureus (MRSA) are significant healthcare-associated pathogens. We sought to identify factors that could be used to predict which patients carry or are infected with VRE or MRSA on admission so that we could obtain cultures selectively from high-risk patients on our burn-trauma unit. We conducted a case-control study of patients admitted to our burn-trauma unit from September 2000 to March 2005 who were colonized or infected with either VRE or MRSA (cases) and patients who were not colonized or infected with one of these organisms (controls). We used logistic regression to construct a model that we subsequently validated based on data collected prospectively from patients admitted from September 2006 to August 2007. In the case-control study, colonization or infection with MRSA or VRE on admission were independently associated with the total days of antimicrobial treatment, age, prior hospitalization, prior operations, and admitting diagnosis (admission for a burn injury was protective). In the cohort study, a prior hospitalization with a length of stay>or=7 days and operations within the past 6 months were significantly associated with colonization or infection on admission. The latter model was 59.3% sensitive. If, we used this model to identify which patients should be cultured on admission, we would have missed 24 (39.3%) of the colonized or infected patients. These patients would not have been placed in isolation (434 missed isolation days, 71.0%) and may have been the source of transmission to other patients. Our model lacked the sensitivity to identify patients colonized or infected with VRE or MRSA. We recommend that units, which care for patients who are at high risk of hospital-acquired infection and having prevalence and transmission rates of VRE or MRSA similar to those in our study, screen all patients for these organisms on admission to the unit.