In March 1978, a strain of methicillin-resistant Staphylococcus aureus was introduced from the community into a university hospital. Within 6 months of admission of the index case, methicillin-resistant 5. aureus was isolated from 30 additional patients, 22 of whom were epidemiologically linked by a common phage type (6/47/54/75/83A) and roommate-to-roommate spread. Sixteen of 31 cases were infected, six with bacteremia. Patients with infections received cephalosporins more frequently before infection than did control subjects (p < 0.05). Patients acquiring methicillin-resistant S. aureus in the intensive care unit had a longer mean stay, had higher overall mortality, and received nafcillin and aminoglycosides more frequently than did cohorted control subjects. By mid-1979, methicillin-resistant S. aureus accounted for 38%, 31%, and 24% of all nosocomial S. aureus postoperative wound, pulmonary, and bloodstream infections, respectively. In hospitals with significant methicillin-resistant 5. aureus isolation rates, initial empiric therapy of presumed S. aureus infection with vancomycin seems warranted.