Critically ill patients, eligible for admission into intensive care units (ICUs), are often hospitalized in other wards due to a lack of ICU beds. Differences in morbidity between patients managed in ICUs and elsewhere are unknown, specifically the morbidity related to hospital-acquired infection. Patients fitting ICU admission criteria were identified by screening five entire hospitals on four separate days. Hospital infections within a 30-day follow-up period were compared in ICU patients and in patients on other wards using Kaplan-Meier curves. Residual differences in the patients' case mix between ICUs and other wards were adjusted for utilizing multivariate Cox models. Of 13415 patients screened, 668 were critically ill. The overall infection rates (per 100 patient-days) were 1.2 for bloodstream infection (BSI) and 1.9 for urinary tract infection (UTI). The adjusted hazard ratios in ICU patients compared with patients on regular wards were 3.1 (P<0.001) for BSI and 2.5 (P<0.001) for UTI. This increased risk persisted even after adjusting for the disparity in the number of cultures sent from ICUs compared with ordinary wards. No interdepartmental differences were found in the rates of pneumonia, surgical wound infections and other infections. Minimizing the differences between characteristics of patients hospitalized in ICUs and in other wards, and controlling for the higher frequency of cultures sent from ICUs did not eliminate the increased risk of BSI and UTI associated with admission into ICUs.