In hospital outbreaks of methicillin-resistant Staphylococcus aureus (MRSA) many patients are initially colonized without infection. The reasons why some progress to infection while others do not are not known. A cohort of 479 hospital patients, initially only colonized with MRSA, was followed prospectively for the development of MRSA infection. Risk factors for progression to infection were assessed using Cox proportional hazards survival analysis. Fifty-three patients (11.1%) developed 68 MRSA infections. Intensive care setting, administration of three or more antibiotics, ulcers, surgical wounds, nasogastric or endotracheal tubes, drains, and urinary or intravenous catheterization were all associated with increased rates of MRSA infection. Multivariate analysis showed that intensive care patients, compared with medical patients, had a higher rate of developing MRSA infection within the first four days of admission, with a hazard ratio of 26.9 (95% CI 5.7-126). Surgical wounds, pressure ulcers and intravenous catheterization were also independent risk factors, with hazard ratios (and 95% CI) of 2.9 (1.3-6.3); 3.0 (1.6-5.7) and 4.7 (1.4-15.6), respectively. These findings suggest that, during an MRSA outbreak, clinical infection would be reduced if surgical and intensive care patients received priority for the prevention of initial colonization with MRSA. Prevention of pressure ulcers, and strict aseptic care of intravenous catheters and surgical wounds would also reduce the development of MRSA infection. Since early treatment with vancomycin is known to reduce the mortality, patients colonized with MRSA who also have one or more of these risk factors may warrant empirical vancomycin therapy at the earliest suggestion of infection.