OBJECTIVE: Nosocomial infections remain a major problem in intensive care units. Several authorities have recommended housing patients in single rooms to prevent cross-transmission of potential pathogens, but this issue is currently debated. The aim of the present study was to compare the rate of nosocomial cross-contamination between patients hosted in single rooms versus bay rooms. DESIGN: Prospective observational data acquisition over 2.5 years. SETTING: A 14-bed medico-surgical ICU, composed of six single-bed rooms plus a six-bed and a two-bed bay room served by the same staff. PATIENTS AND PARTICIPANTS: All patients admitted from 1 July 2002 to 31 December 2004. INTERVENTIONS: None. MEASUREMENTS AND RESULTS: Prevalence of methicillin-resistant Staphylococcus aureus (MRSA) in admitted patients was 1.1% and acquisition rate 2.4%. The incidence density of MRSA acquisition was 4.1 [95% CI 2.7-6.3]/1,000 patient-days in bay rooms versus 1.3 [0.5-3.4]/1,000 patient-days in single rooms (p<0.001). Pseudomonas spp. acquisition rate was 3.9 [2.5-6.1]/1,000 patient-days in bay rooms versus 0.7 [0.2-2.4]/1,000 patient-days in single rooms (p<0.001), and Candida spp. colonization was 38.4 [33.3-44.1]/1,000 patient-days in bay rooms versus 13.8 [10.2-18.6]/1,000 patient-days (p<0.001). By multivariate analysis, the relative risk of MRSA, Pseudomonas aeruginosa and Candida spp. acquisition in single rooms or cubicles versus bay rooms was 0.65, 0.61 and 0.75 respectively. CONCLUSIONS: These data suggest that in an institution where MRSA is not hyperendemic, infection control measures may be more effective to prevent cross-transmission of microorganisms in patients housed in single rooms.