The transfusion of blood products, especially red cell concentrates, in critically ill patients is controversial and benefits of red cell concentrate transfusion in these patients have not been clearly demonstrated. We performed a prospective observational study to compare best evidence to actual practice of red cell concentrate and other blood product administration in an intensive care unit (ICU) in a university-associated tertiary hospital. All primary admissions during a 28-day period were included in the study and data collected included transfusion of red cells and blood products, patient demographics and ICU and hospital outcome. One hundred and seventy-five admissions were studied; 44% followed cardiac surgery. Forty-one patients (23%) received red cell concentrates in ICU, with 120 units transfused in 61 separate episodes. Other blood product usage was minimal. One third (20/61) of red cell concentrate transfusion episodes were of a single unit. The mean (+/- SD) pre-transfusion haemoglobin was 7.9 +/- 1.1 g/dl. Despite transfusion, such patients left ICU with a lower haemoglobin concentration compared with untransfused ICU patients (9.5 +/- 1.0 versus 10.5 +/- 2.1 g/dl; P < 0.001). Cardiac surgical patients received similar red cell transfusion to general ICU patients. Univariate analysis showed no significant difference in mortality between patients who did or did not receive red cell concentrate transfusion (P = 0.17). However, red cell concentrate transfusion was associated with a reduced adjusted mortality both in ICU (OR 0.13, 95% CI 0.02-0.73) and in hospital at 28 days (OR 0.10, 95% CI 0.02-0.58). The low red cell concentrate and blood product usage in our ICU were consistent with restrictive transfusion practice and selective red cell concentrate transfusion was associated with reduced mortality.