BACKGROUND Liver transplantation from donors after cardiac death (DCD) resolves donor shortages. OBJECTIVE We investigated the optimal time for subnormothermic oxygenated perfusion in DCD liver transplantation. METHODS Ten F1 pigs (body weight: 27-32 kg) were allocated to 2 groups: the heart beating group (n = 6), from which livers were retrieved while the heart was beating, and the donation after cardiac death (DCD) group (n = 4), in which liver retrieval was performed on pigs under apnea-induced cardiac arrest for 20 minutes. In both groups, the livers were kept in cold storage for 2 hours after retrieval and perfused with a subnormothermic oxygenated Krebs-Henseleit buffer for 120 minutes. We used a novel perfusion device, which can set maximum perfusion pressures of arteries and portal vein, developed by Asahikawa Medical University and Chuo Seiko Co. Bile production, liver enzymes, and inflammatory cytokines were measured and the sinusoidal space, using tissue specimens taken from liver grafts, was measured at 30, 60, 90, and 120 minutes after the start of perfusion. RESULTS Bile production peaked at 90 minutes. Significantly higher levels of liver enzymes and inflammatory cytokines were found in the DCD group (P < .05). The release of liver enzymes peaked at 60 minutes and that of inflammatory cytokines peaked at 90 minutes. The hepatic sinusoidal space was wide at 90 minutes and narrowed after 120 minutes. CONCLUSIONS The results suggest that subnormothermic oxygenation perfusion may maintain optimal graft condition until around 90 minutes and perfusion for more than 120 minutes may be counterproductive.