From January to December 1994, 752 consecutive patients admitted to intensive care units (ICU) for more than two days were studied prospectively for Staphylococcus aureus colonization and infection. Nasal swabs were obtained at admission and weekly during the ICU stay. At ICU admission 166 patients (22.1%) were Staphylococcus aureus nasal carriers, while 586 were free of nasal colonization. Of the 166 nasal carriers, 163 harbored methicillin-sensitive Staphylococcus aureus (MSSA) and three methicillin-resistant Staphylococcus aureus (MRSA). During the ICU stay 24 of the 586 noncolonized patients became nasal carriers (11 MSSA and 13 MRSA), and one nasal carrier initially colonized by MSSA was reconlonized by MRSA. Staphylococcal infections were documented in 51 (6.8%) of the total 752 patients. After 14 days of ICU stay, the probability of developing staphylococcal infections was significantly higher for those patients who were nasal carriers at ICU admission than for those found to be initially negative (relative risk 59.6, 95% CI 20.37-184.32; p < 0.0001). In patients with ICU-acquired nasal colonization, most infections were documented prior to or at the time of the detection of the nasal colonization; thus, in this group of patients nasal carriage showed a lower predictive value for subsequent Staphylococcus aureus infections that that described classically. Paired isolates of nasal colonizing and clinical strains were studied by pulsed-field gel electrophoresis (PFGE) and mecA polymorphism analysis in 30 patients; identity was demonstrated in all but two patients. The results suggest that, outside the setting of an outbreak of MRSA, the detection of Staphylococcus aureus nasal carriers on admission may be particularly useful in identifying those patients who are at high risk for developing staphylococcal infections during their ICU stay.