Blood transfusion has been associated with infection; however, the collinearity of injury severity has not been clearly addressed to show a direct relationship. Using more rigorous analysis, we aimed to untangle the effect of injury severity from transfusion leading to sepsis. We hypothesized that blood transfusion independently increases infection in massively transfused versus nontransfused patients with matched Injury Severity Scores (ISSs). We performed a matched cohort study measuring infection rates in trauma patients receiving massive transfusion. Control subjects were contemporaneous patients with matched ISS receiving no blood. Infection was defined as intraperitoneal or intrathoracic abscesses, pneumonia, urinary tract infection, or bacteremia. Multivariate logistic and univariate analysis was completed. Infection rate was 61 per cent in 44 transfused patients versus 20 per cent in 44 control subjects (P = 0.001). Odds of infection were eightfold greater in transfused patients (OR, 7.97; 95% CI, 2.3 to 27.5; P < 0.001) independent of ISS, Glasgow Coma Scale, mechanism, and age. Infection was most associated with transfusion of packed red blood cells (PRBCs), although transfusion of other blood products had strong collinearity with PRBCs. Transfused patients had eight times the risk of infection independent of ISS; this appears to be the result of PRBC transfusion. Modifying the ratio of components in transfusion protocols favoring plasma may cause less infection after injury.