Methicillin-resistant Staphylococcus aureus (MRSA) is a major cause of health care-associated infections. Multiple factors, including transmission from unrecognized reservoirs of MRSA, are responsible for failure to control the spread of MRSA. We conducted prospective surveillance to determine the frequency of gastrointestinal colonization with MRSA among patients and its possible impact on nosocomial transmission of MRSA. Stool specimens submitted for Clostridium difficile toxin A/B assays were routinely inoculated on colistin-naladixic acid agar plates, and S. aureus was identified by using standard methods. Methicillin resistance was confirmed by growth on oxacillin-salt screening agar. For patients whose stool yielded MRSA, information regarding any previous cultures positive for MRSA or other organisms that would require contact precautions was obtained from the laboratory's computer system. During a 1-year period, 151 (9.8%) of 1,543 patients who had one or more stool specimens screened had MRSA in their stool. Ninety-three (62%) of the 151 patients had no previous history of MRSA colonization or infection. Of these 93, 75 were inpatients. Sixty (80%) of the 75 inpatients with no previous history of MRSA were not under "contact precautions." The 60 patients would have spent an estimated total of 267 days without being placed under contact precautions if their positive stool cultures had not resulted in their being isolated. Placing patients under contact precautions based on their positive stool cultures prevented an estimated 35 episodes of MRSA transmission. We conclude that gastrointestinal colonization with MRSA may serve as an unrecognized reservoir from which transmission of MRSA may occur in health care facilities.