To measure the prevalence of, and to establish predictors for, the nasal carriage of methicillin-resistant Staphylococcus aureus (MRSA) at hospital admission. To evaluate mannitol-salt agar with oxacillin for the simultaneous detection and identification of MRSA from nasal swabs. Three-month prospective case-control survey, with data collected from interviews and computerized databases. The criterion standard for MRSA detection was culture on Mueller-Hinton agar with oxacillin 6 microg/mL (National Committee for Clinical Laboratory Standards method). 320-bed tertiary-care hospital. 387 patients screened within 24 hours after admission, including 10 MRSA carriers (cases), 291 patients with no S aureus, and 86 patients with methicillin-susceptible S aureus. The prevalence of MRSA nasal carriage was 2.6%, whereas the prevalence of carriage was 3.1% when both nasal and wound cultures were performed. The significant predictors of carriage were a prior detection of MRSA, open wounds, diabetes mellitus, treatments by injection, prior nursing home stays, visits at home by a nurse, and prior antibiotic treatments. Cases had stayed for longer periods in hospitals and had received longer antibiotic treatments within a year. Eighty patients (including the 10 cases) had diabetes, had been exposed to healthcare facilities within a year, and had antibiotics within 6 months. The sensitivity and negative predictive value of nasal swabs on mannitol-salt agar with oxacillin were 60% and 71%, respectively. MRSA carriage on admission to the hospital may be an increasing and underestimated problem. Further studies are needed to develop and validate a sensitive and specific prediction rule.