To assess the severity of ischemic liver injury, we examined release of mitochondrial aspartate aminotransferase (EC 18.104.22.168) and its cytoplasmic isozyme from the ischemic rat liver into the circulation. Their patterns of leakage were quite different: the level of cytoplasmic aspartate aminotransferase reached a peak soon after the circulation to the ischemic liver was restored, while that of mitochondrial aspartate aminotransferase increased slowly, reaching a maximum after more than 10 hr. On anoxic incubation of mitochondria isolated from the normal liver, oxidative phosphorylation capacity was lost within 2 hr, at which time no leakage of matrix enzymes was observed: more than 10 hr after-loss-of-oxidative phosphorylation were needed for the matrix enzymes to leak out of the mitochondrial membrane. Since the viability of cells is considered to depend on the capacity of oxidative phosphorylation, it is highly likely that the delayed appearance of mitochondrial aspartate aminotransferase in blood indicates the postmortem changes of injured cells. In fact, the cumulative activity of mitochondrial aspartate aminotransferase but not cytoplasmic aspartate aminotransferase in circulation after ischemic liver injury correlated fairly well with the decrease of total adenine nucleotides which were monitored to measure viable cells. The difference between mitochondrial aspartate aminotransferase and cytoplasmic aspartate aminotransferase as quantitative indices of hepatic necrosis may be due to the relative stability of the former and significant inactivation of the latter during hepatic ischemia. Therefore, the determination of mitochondrial aspartate aminotransferase in blood may be useful in the assessment of liver necrosis after ischemic injury.