All episodes of ventilator-associated pneumonia (VAP) caused by Staphylococcus aureus were prospectively analyzed for a 30-mo period. Methicillin-sensitive S. aureus (MSSA) was isolated in 38 episodes and methicillin-resistant S. aureus (MRSA) in 11 others. The two groups were similar regarding sex, severity of underlying diseases, prior surgery, and presence of renal failure, diabetes, cardiopathy, and coma. MRSA-infected persons were more likely to have received steroids before developing infection (relative risk [RR] = 3.45, 95% confidence interval [CI] = 1.38-8.59), to have been ventilated > 6 d (RR = 2.03, 95% CI = 1.36-3.03), to have been older than 25 yr (RR = 1.50, 95% CI = 1.09-2.06), and to have had preceding chronic obstructive pulmonary disease (RR = 2.76, 95% CI = 0.89-8.56) than MSSA-infected patients. MSSA-infected persons were more likely than MRSA-infected patients to have cranioencephalic trauma (RR = 1.94, 95% CI = 1.22-3.09). All patients with MRSA VAP had previously received antibiotics, compared with only 21.1% of those with MSSA infection (p < 0.000001). The incidence of empyema was similar in both groups; nevertheless, the presence of bacteremia and septic shock was more frequent in the MRSA group. Finally, mortality directly related to pneumonia was significantly higher among patients with MRSA episodes (RR = 20.72, 95% CI = 2.78-154.35). This analysis was repeated for monomicrobial episodes, and the difference remained statistically significant. We conclude that MRSA and MSSA strains infect patients with different demographic profiles; previous antibiotic therapy is the most important risk factor for developing MRSA infection.(ABSTRACT TRUNCATED AT 250 WORDS)