Over the 5-year period from 1990 to 1994, a prospective cohort study was conducted to define the clinical and epidemiological characteristics of ventilator-associated methicillin-resistant Staphylococcus aureus (MRSA) pneumonia acquired during a large-scale outbreak of MRSA infection. Of 2411 mechanically ventilated patients, 347 (14.4%) acquired MRSA, 220 (63.4%) had MRSA positive respiratory tract samples and 41 (18.6%) developed ventilator-associated MRSA pneumonia. The overall attack rate for ventilator-associated MRSA pneumonia was 1.56 episodes/1000 ventilator days, but annual attack rates varied according to the trend of the outbreak (range 4.9-0.2). In comparison with methicillin-sensitive Staphylococcus aureus (MSSA), which was implicated in 98 episodes of ventilator-associated pneumonia, MRSA caused exclusively late-onset ventilator-associated pneumonia, while MSSA caused both early-onset [55 of 98 (56.1%) episodes] and late-onset [43 of 98 (43.8%) episodes] ventilator-associated pneumonia. Logistic regression analysis of all patients with Staphylococcus aureus pneumonia revealed intubation for more than 3 days (odds ratio (OR),1.11; confidence interval (CI):1.03-1.18) and prior bronchoscopy (OR,5.8; CI,1.85-18.19) to be independent variables associated with MRSA pneumonia. The results indicate that MRSA ventilator-associated pneumonia is a frequent complication in intensive care patients, manifesting itself as late-onset pneumonia in patients who have been intubated for prolonged periods and/or have often undergoing previous bronchoscopy.