Fluid resuscitation is considered to be an integral component of the management of hemorrhagic shock. Numerous experimental studies of hypovolemic shock have confirmed the value of volume infusions, but in these models the rate, volume, and duration of bleeding are carefully controlled. The results of such studies may not be applicable to clinical hemorrhage, in which bleeding continues unabated. Male Sprague-Dawley rats weighing 250 to 390 g were anesthetized, and a femoral artery and vein were cannulated for constant blood pressure monitoring and fluid infusion. Through a midline abdominal incision, the distal ileocolic artery and vein were transected and allowed to bleed freely into the peritoneal cavity. The abdomen was closed and the animals were randomized to one of five groups: no resuscitation; small volume lactated Ringer's solution; large volume lactated Ringer's; small volume hetastarch; or large volume hetastarch. After 3 hours or at spontaneous death, blood was withdrawn to measure hematocrit, platelet count, and fibrinogen. Blood in the peritoneal cavity was collected and measured. Animals that received either lactated Ringer's or hetastarch had more bleeding into the peritoneal cavity and a greater dilution of clotting factors than animals that received no resuscitation fluids (P < 0.05). In addition, survival was highest in unresuscitated animals, although only the small volume hetastarch group had a significantly lower survival when independently compared with no resuscitation (P < 0.05). These results suggest that in traumatic shock, fluid resuscitation should be minimized until mechanical control of bleeding can be achieved.