To describe methicillin-resistant Staphylococcus aureus (MRSA) control in a hospital, including a surgical intensive care unit (SICU) outbreak.Prospective surveillance of newly identified patients with MRSA. Barrier isolation (disposable gloves for direct contact with patient or immediate environment) was used for the routine care of hospitalized MRSA patients as of October 1991. Beginning in 1992, MRSA isolates were typed by restriction endonuclease enzyme analysis of plasmid DNA (REAP) and/or pulsed-field gel electrophoresis of genomic DNA (PFGE). Surveillance information and MRSA typing were used concurrently to identify nosocomial case clustering, confirm cross-infection, and support a need for additional outbreak control interventions.University-affiliated public hospital.Patients with newly identified MRSA colonization or infection from 1991 through 1993 and epidemiologically associated staff providing care to eight SICU patients in an outbreak.Barrier isolation for affected and unaffected patients in and admitted to the SICU institution when the outbreak was identified and cross-infection confirmed. Anterior nares cultures of staff in contact with outbreak cases for detection of MRSA colonization.Fifty-six hospitalized patients with community-acquired MRSA and 80 patients with nosocomial MRSA colonization or infection were identified during the 3 years. After the introduction of barrier isolation, the annual frequency of new nosocomial MRSA cases decreased and only one outbreak (eight cases in the SICU) caused by type-related isolates occurred. The other 35 nosocomial cases of MRSA during 1992 and 1993 were not epidemiologically related or were caused by isolates with different types. The SICU outbreak ended after instituting barrier isolation for all patients (with and without MRSA) in and admitted to the unit. Six colonized SICU staff were identified. All outbreak cases had identical or related MRSA types by PFGE and REAP. Staff isolates were different from case isolates by typing, and staff were not restricted and not given treatment for colonization. After more than 6 months of follow up, no further outbreaks of MRSA in the SICU or elsewhere in the hospital occurred despite returning to barrier isolation for affected patients only.MRSA in hospitals and outbreaks of MRSA in ICUs can be controlled by surveillance and minimal barrier interventions. REAP or PFGE typing of MRSA can be used to support or refute the presence of cross-transmission. Typing also may be helpful when planning and assessing the effectiveness of interventions directed at endemic, as well as outbreak, MRSA control.