The methods of hemostasis used for liver injuries were evaluated prospectively in 637 patients treated at Detroit General Hospital during a 5-year period. Variables evaluated included severity of injury, presence or absence of bleeding, and methods of hemostasis, The liver injury was either not bleeding or was controlled by temporary pack compression during laparotomy in 325 patients: none of these patients, including the 284 in whom no hemostatic procedure was used, rebled postoperatively. Active bleeding at laparotomy was directly related to the severity of liver injury, and required some hemostatic procedure in 312 patients. The methods of hemostasis were liver sutures (244 patients), nonanatomic resection (30 patients), anatomic resection (21 patients), hepatic artery ligation (nine patients), hepatotomy with intraparenchymal vascular control (five patients), and temporary internal pack with later re-operation (three patients). Rebleeding occurred in eight of the 243 patients who survived (seven after liver sutures and one after nonanatomic resection) and four required re-operation for control of bleeding. Sixty-nine patients with active bleeding died. Death on the table in 38 patients was related primarily to uncontrolled bleeding from liver and major vessel injury. Postoperative rebleeding from the liver occurred in 14 of 31 patients who died after surgery: following initial control by liver sutures (seven patients); anatomic resection (four patients); and hepatic artery ligation (three patients). There was no apparent relationship between any hemostatic procedure and the subsequent appearance of the hepatic ischemia or parahepatic abscess. Based on this experience, the merits and detriments of individual hemostatic procedures are presented.