It has been independently postulated that nutritional status is a modulator of the hepatic injury response to hypoxia and that glucose may be a poor substrate for hepatocellular metabolism. This study provides data linking these two concepts within the framework of metabolic zonation of the liver. With the use of a hypoxically perfused isolated rat liver model, cellular injury, as reflected by aspartate aminotransferase (AST) release, was significantly greater in the liver of fasted (mean AST 489 U/g liver at 3 h) than fed (40 U/g) animals. The extent of injury during hypoxia was decreased to a comparable degree in fasted livers perfused with Wisconsin solution (27 U/g) or 20 mM fructose (51 U/g). Perfusion with (11.5 mM) glucose plus insulin provided no hepatoprotection (791 U/g); however, supraphysiological amounts of glucose (100 mM) with (310 U/g) or without (321 U/g) insulin (10 U) or dihydroxyacetone (220 U/g) provided a modest reduction in AST release. Cellular injury measured by trypan blue uptake showed a marked zonal pattern, with upstream regions incurring greater parenchymal and nonparenchymal injury than downstream areas. These data that indicate that exogenous glucose is poorly utilized as an energy substrate by the liver during hypoxia are consistent with data from the fasted-refed rat model, suggesting a "glucose paradox" in the liver. The findings also suggest that low levels of oxygen are an important factor mediating "hypoxic" liver injury.