OBJECTIVE To evaluate the clonality of methicillin-resistant Staphylococcus aureus (MRSA) strains among hospitalized patients. METHODS University-affiliated, 465-bed tertiary-care teaching hospital with adjacent cancer clinic in Hamilton, Ontario, Canada. METHODS Thirty-five colonized and 30 infected patients from January 2, 1992, through August 31, 1993, were investigated retrospectively. Analysis by restriction fragment-length polymorphisms of ribosomal RNA genes (ie, ribotyping) of 103 nosocomial isolates of MRSA from these 65 patients and of 25 selected unrelated strains was completed. Ribotyping results were compared with the phage typing data obtained prospectively during the course of prospective MRSA surveillance. RESULTS HindIII ribotyping was more discriminating than phage typing when epidemiologically unrelated strains were differentiated by these methods (19 different ribotypes versus 14 page types; P < .005). Two early index cases were identified. Isolates from the index cases were two different strains, identified by ribotyping analysis as ribotype A (clonal group 1) and ribotype B (clonal group 2), respectively. These two ribotypes were not found when typing the unrelated control strains. Thirty-six colonized and infected patients (55%) had clonal group 1 isolates, and 20 (31%) had clonal group 2 isolates. These two clones emerged in the hospital in January and February 1992 and dominated the entire investigated period. There also were six patients with an additional clonal group (group 4) that emerged and disappeared in the second quarter of 1993. CONCLUSIONS This study highlights the utility of ribotyping in investigating nosocomial MRSA. Three MRSA clones caused nosocomial colonization or infection in patients at this hospital. Two of these MRSA clones, once introduced, were maintained among our patients throughout the study period.