Contamination of blood by bowel contents has been generally assumed as an absolute contraindication to autotransfusion. Since abdominal trauma is frequently accompanied by bowel injury and massive blood loss, a potential major use for autotransfusion has thus been precluded. To test this presumption, autologous blood grossly contaminated with feces was incubated in the peritoneal cavity and then autotransfused in dogs. The animals were hemorrhaged 20, 30, or 40% of their estimated blood volume, producing mild to severe hypovolemic shock. Reinfusion of contaminated blood had little effect on survival with 20 or 30% hemorrhage, but contamination markedly decreased survival with 40% hemorrhage:90% survived without contamination while only 30% survived with contamination. The use of antibiotics in a similar group of dogs subjected to 40% hemorrhage essentially eliminated the risk of autotransfusion: 90% of these dogs survived autotransfusion of contaminated blood.