CONTEXT: Severe, uncompensated, traumatic hemorrhagic shock causes significant morbidity and mortality, but resuscitation with an oxygen-carrying fluid might improve patient outcomes. OBJECTIVE: To determine if the infusion of up to 1000 mL of diaspirin cross-linked hemoglobin (DCLHb) during the initial hospital resuscitation could reduce 28-day mortality in traumatic hemorrhagic shock patients. DESIGN AND SETTING: Multicenter, randomized, controlled, single-blinded efficacy trial conducted between February 1997 and January 1998 at 18 US trauma centers selected for their high volume of critically injured trauma patients, but 1 did not enroll patients. PATIENTS: A total of 112 patients with traumatic hemorrhagic shock and unstable vital signs or a critical base deficit, who had a mean (SD) patient age of 39 (20) years. Of the infused patients, 79% were male and 56% were white. An exception to informed consent was used when necessary. INTERVENTION: All patients were to be infused with 500 mL of DCLHb or saline solution. Critically ill patients who still met entry criteria could have received up to an additional 500 mL during the 1-hour infusion period. MAIN OUTCOME MEASURES: Twenty-eight day mortality, 28-day morbidity, 48-hour mortality, and 24-hour lactate levels. RESULTS: Of the 112 patients, 98 (88%) were infused with DCLHb or saline solution. At 28 days, 24 (46%) of the 52 patients infused with DCLHb died, and 8 (17%) of the 46 patients infused with the saline solution died (P = .003). At 48 hours, 20 (38%) of the 52 patients infused with DCLHb died and 7 (15%) of the 46 patients infused with the saline solution died (P = .01). The 28-day morbidity rate, as measured by the multiple organ dysfunction score, was 72% higher in the DCLHb group (P = .03). There was no difference in adverse event rates or the 24-hour lactate levels. CONCLUSIONS: Mortality was higher for patients treated with DCLHb. Although further analysis should investigate whether the mortality difference was solely due to a direct treatment effect or to other factors, DCLHb does not appear to be an effective resuscitation fluid.