Over a 54-hour period, blood was removed from 8 adult sheep (body weight, 38.1 +/- 0.5 kg mean +/- SEM) in 9 episodes, 5 on day 1, 3 on day 2, and 1 on day 3. Cumulative blood loss was 1,630 +/- 63, 2,380 +/- 71, and 2,693 +/- 69 ml on days 1, 2, and 3, respectively. Blood samples (20 ml) were collected from 5 control ewes (33.8 +/- 2.8 kg) at equivalent times. Over the first day, mean arterial blood pressure decreased in the hemorrhaged sheep from 101 +/- 2 mm of Hg to 76 +/- 5 mm of Hg, but returned to control values by the beginning of the second day and, thereafter, was not different from control values. Heart rate was increased after the first hemorrhage episode and remained high throughout the entire protocol. Over the entire period, there were statistically significant decreases in hematocrit, plasma osmolality, sodium, total calcium (P < 0.001), potassium, and chloride values (P < 0.05). There was no change in plasma phosphate, bicarbonate, creatinine, or magnesium concentrations and an increase in plasma urea nitrogen (P < 0.001) concentrations. Plasma arginine vasopressin concentration was increased significantly (P < 0.001) over the entire period. Plasma ACTH concentration was significantly (P < 0.05) increased over time, but only some values on day 1 were significantly outside the normal range of the control group data. Because of wide variation between sheep, the group data for aldosterone were not significantly different from control values. Blood volume was restored on day 1 with fluid of osmolality, Na, and Cl composition equivalent to that of plasma. The effects of arginine vasopressin were apparent by day 2, when the major decrease in osmolality and Na and Cl concentrations were observed The sheep has good capacity to withstand severe, prolonged hemorrhage, most likely because of a large reserve of RBC in the spleen; hematocrit remained at 31% of control values when an estimated 100% of initial circulating blood volume had been removed.