Divergent views remain regarding the safety of treating anemia with red blood cell (RBC) transfusion in patients with acute coronary syndrome (ACS). We used a prospective database to study effect of RBC transfusion in patients with acute myocardial infarction (MI; n = 2,358). Cox regression models were used to determine the association between RBC transfusion and 6-month outcomes, incorporating transfusion as a time-dependent variable. The models adjusted for baseline variables, propensity for transfusion, and nadir hemoglobin previous to the transfusion. One hundred ninety-two patients (8.1%) received RBC transfusion. Six-month mortality rates were higher in patients receiving transfusion (28.1% vs 11.7%, p <0.0001). The adjusted hazard ratio (HR) for mortality was 1.9 in transfused patients (95% confidence interval [CI] 1.3 to 2.9). Interaction between RBC transfusion and nadir hemoglobin with respect to mortality (p = 0.004) was significant. Stratified analyses showed a protective effect of transfusion in patients with nadir hemoglobin < or=8 g/dL (adjusted HR 0.13, 95% CI 0.03 to 0.65, p = 0.013). By contrast, transfusion was associated with increased mortality in patients with nadir hemoglobin >8 g/dL (adjusted HR 2.2, 95% CI 1.5 to 3.3; p <0.0001). Similar results were obtained for the composite end point of death/MI/heart failure (p for interaction = 0.04). In conclusion, RBC transfusion in patients with acute MI and hemoglobin < or =8 g/dL may be appropriate. The increased mortality observed in transfused patients with nadir hemoglobin above 8 g/dL underscores the clinical difficulty of balancing risks and benefits of RBC transfusion in the setting of ACS.