Bloodstream infection (BSI) is a serious complication of critical illness but it is uncertain whether acquisition of BSI in the intensive care unit (ICU) increases the risk of death. A study was conducted among all Calgary health region (population approximately 1 million) adults admitted to ICUs for 48 h or more during a three-year period to investigate the occurrence, microbiology and risk factors for developing an ICU-acquired BSI and to determine whether these infections independently predict mortality. One hundred and ninety-nine ICU-acquired BSI episodes occurred during 4933 ICU admissions for a cumulative incidence of 4% and an incidence density of 5.4 per 1000 ICU days. The most common isolates were Staphylococcus aureus (18%), coagulase-negative staphylococci (11%), and Enterococcus faecalis (8%); 12% of infections were due to antimicrobial-resistant bacteria. Admission to the regional neurosurgery/trauma ICU [odds ratio (OR) 2.86; 95% confidence interval (CI) 2.10-3.90] and increasing Acute Physiology and Chronic Health Evaluation II (APACHE II) score (OR 1.05 per point, 95% CI 1.03-1.07) were associated with higher risk, whereas a surgical diagnosis (OR 0.69; 95% CI 0.52-0.93) was associated with lower risk of developing ICU-acquired BSI in logistic regression analysis. The crude in-hospital death rate was 45% for patients with ICU-acquired BSI compared with 21% for those without (P