All patients positive for methicillin-resistant Staphylococcus aureus (MRSA) at the University Hospitals of Geneva, Switzerland, between 1989 and 1997 (N = 1771) were included in a cohort study to evaluate the consequences of delayed containment of a hospital-wide outbreak occurring during a 4-year absence of MRSA control measures. The effects of efforts to control both the MRSA reservoir and the number of bacteraemic patients were assessed. Intensive infection control measures were initiated in 1993 and included patient screening, on-site surveillance, contact isolation, a computerized alert system, and hospital-wide promotion of hand hygiene. An increase in the rate of new MRSA-infected or -colonized patients was observed between 1989 and 1994 (from 0.05 to 0.60 cases per 100 admissions), which subsequently decreased to 0.24 cases in 1997 (P<0.001). However, the proportion of laboratory-documented methicillin-resistant isolates among all S. aureus showed little variation in the years from 1993 onwards (range, 19-24%), reflecting the result of an increase in the number of screening cultures. The annual number of patients with MRSA bacteraemia strongly correlated with the hospital-wide prevalence of MRSA patients (R(2)= 0.60; P = 0.01) and the rate of new MRSA patients (R(2)= 0.97; P<0.001). Consequently, the attack rate of nosocomial MRSA bacteraemia served as an excellent marker for the MRSA patient reservoir. In conclusion, despite delayed implementation, infection control measures had a substantial impact on both the reservoir of MRSA patients and the attack rate of MRSA bacteraemia.