In 284 US intensive care units the CRIT study (Anemia and blood transfusion in the critically ill--Current clinical practice in the United States) assessed allogeneic red blood cell (RBC) transfusion and outcome in 4892 patients. As in the former European ABC study (Anemia and blood transfusion in the critically ill), the mean pretransfusion hemoglobin was approximately 8.5 g/dl and RBC transfusions were independently associated with an increased mortality. These studies were purely observational and, therefore, despite the finest statistical models indicating that RBC transfusions were independently associated with a higher mortality, it remains possible that this adverse outcome is not due to a harmful effect of RBC transfusion in itself, but merely reflects the fact that transfused patients were sicker to start with. The definitive call is still out, but one mechanism by which RBC transfusion might be harmful now appears less likely; namely, storage lesion. In the CRIT study, mortality was not increased in patients receiving 'old' RBCs (>14 days stored) versus 'fresh' RBCs. The effect of leukoreduction could not be assessed since mainly nonleukoreduced RBCs were transfused. The evidence is mounting, however, that RBC transfusions are efficacious only when oxygen delivery is compromised. What can be done to diminish the use of RBC transfusions, its costs and side effects in intensive care medicine? There are two important options available today: decreasing blood loss for diagnostic purposes using pediatric sampling tubes, and establishing restrictive multidisciplinary transfusion guidelines and implementing them in daily clinical practice.