Background: Trauma is the leading cause of death and disability in patients aged 1–46 y. Severely injured patients experience considerable blood loss and hemorrhagic shock requiring treatment with massive transfusion of red blood cells (RBCs). Preclinical and retrospective human studies in trauma patients have suggested that poorer therapeutic efficacy, increased severity of organ injury, and increased bacterial infection are associated with transfusion of large volumes of stored RBCs, although the mechanisms are not fully understood. Methods and findings: We developed a murine model of trauma hemorrhage (TH) followed by resuscitation with plasma and leukoreduced RBCs (in a 1:1 ratio) that were banked for 0 (fresh) or 14 (stored) days. Two days later, lungs were infected with Pseudomonas aeruginosa K-strain (PAK). Resuscitation with stored RBCs significantly increased the severity of lung injury caused by P. aeruginosa, as demonstrated by higher mortality (median survival 35 h for fresh RBC group and 8 h for stored RBC group; p < 0.001), increased pulmonary edema (mean [95% CI] 106.4 μl [88.5–124.3] for fresh RBCs and 192.5 μl [140.9–244.0] for stored RBCs; p = 0.003), and higher bacterial numbers in the lung (mean [95% CI] 1.2 × 107[−1.0 × 107to 2.5 × 107] for fresh RBCs and 3.6 × 107[2.5 × 107to 4.7 × 107] for stored RBCs; p = 0.014). The mechanism underlying this increased infection susceptibility and severity was free-heme-dependent, as recombinant hemopexin or pharmacological inhibition or genetic deletion of toll-like receptor 4 (TLR4) during TH and resuscitation completely prevented P. aeruginosa–induced mortality after stored RBC transfusion (p < 0.001 for all groups relative to stored RBC group). Evidence from studies transfusing fresh and stored RBCs mixed with stored and fresh RBC supernatants, respectively, indicated that heme arising both during storage and from RBC hemolysis post-resuscitation plays a role in increased mortality after PAK (p < 0.001). Heme also increased endothelial permeability and inhibited macrophage-dependent phagocytosis in cultured cells. Stored RBCs also increased circulating high mobility group box 1 (HMGB1; mean [95% CI] 15.4 ng/ml [6.7–24.0] for fresh RBCs and 50.3 ng/ml [12.3–88.2] for stored RBCs), and anti-HMGB1 blocking antibody protected against PAK-induced mortality in vivo (p = 0.001) and restored macrophage-dependent phagocytosis of P. aeruginosa in vitro. Finally, we showed that TH patients, admitted to the University of Alabama at Birmingham ER between 1 January 2015 and 30 April 2016 (n = 50), received high micromolar–millimolar levels of heme proportional to the number of units transfused, sufficient to overwhelm endogenous hemopexin levels early after TH and resuscitation. Limitations of the study include lack of assessment of temporal changes in different products of hemolysis after resuscitation and the small sample size precluding testing of associations between heme levels and adverse outcomes in resuscitated TH patients. Conclusions: We provide evidence that large volume resuscitation with stored blood, compared to fresh blood, in mice increases mortality from subsequent pneumonia, which occurs via mechanisms sensitive to hemopexin and TLR4 and HMGB1 inhibition.