In-patients at a London hospital over one year from whom the south-east England strain of ‘epidemic’ methicillin-resistant Staphylococcus aureus (MRSA) was isolated were compared with in-patients with strains of methicillin-sensitive Staphylococcus aureus (MSSA). MRSA were virtually entirely hospital-acquired; isolates before 10'days were uncommon and related to recent previous admission. Thereafter first isolates occurred at a fairly constant daily rate of about 1·9 per 1000 in-patients. Acquisition of MSSA after more than 4 days in hospital occurred at a similar constant rate. Such strains were less likely to be penicillin-sensitive than strains isolated in the first 4 days after admission (11 vs. 22%) and were considered to be hospital-acquired. The single MRSA strain caused 40 infections in a year, about half of all hospital-acquired staphylococcal infections. Patients prescribed anti-staphylococcal antibiotics and patients with indwelling cannulae both had about a ninefold increased risk of acquiring MRSA. There was no reciprocal increase in MSSA infections after control measures had substantially reduced the number of MRSA infections.