OBJECTIVE: To study the epidemiology of methicillin-resistant Staphylococcus aureus (MRSA) over a prolonged period of time with the aid of a molecular typing method (ribotyping). SETTING: A 1,000-bed tertiary university medical center. PATIENTS AND METHODS: Defined epidemiological data were recorded for all patients culture-positive for MRSA between 1989 and 1992. Ribotyping of MRSA strains was performed using three restriction enzymes: EcoRV, HindIII, and KpnI. RESULTS: From 1989 to 1992, MRSA was isolated from clinical specimens in 98 patients and from surveillance cultures in 27 patients. Among the 122 isolates available for typing, 26 different ribotypes were identified. In 20% of the cases, MRSA was community-acquired, and a third of these patients never had been hospitalized previously. Nine ribotypes were responsible for more than one case (2 to 64 patients); 17 appeared only once. Epidemiological data correlated with ribotyping results revealed 14 epidemiologic clusters involving six different ribotypes, whereas only three outbreaks were suspected initially. The median follow-up after the last isolation of a given ribotype was 14 months (range, 1 to 42) for clusters and 25 months (range, 1 to 46) for ribotypes that appeared only once. During clusters, only 16% of the cases occurred after the implementation of control measures in the ward (breakthrough cases). CONCLUSIONS: The high diversity of MRSA strains observed over 4 years suggested that new strains were introduced continuously in our hospital. Furthermore, that 17 ribotypes were isolated only once, that breakthrough cases represented only 16% of the cases in clusters, and that the follow-up duration after the last isolation of a given ribotype was more than 14 months suggest that infection control measures were effective in limiting the nosocomial spread of MRSA over a prolonged period of time.