Primary cultures of adult rat hepatocytes were kept for 46 h with either insulin ('insulin cells') or glucagon ('glucagon cells') as the dominant hormone under different oxygen concentrations with 13% (v/v) O2 mimicking arterial and 4% hepatovenous levels. Thereafter metabolic rates were measured for a 2 h period under the same ('overall long-term O2 effects') or a different ('short-term O2 effects') oxygen concentration. From the differences of the two effects the 'intrinsic long-term O2 effects' were derived. Glycolysis, as measured in 'insulin-cells', was stimulated by low O2 levels. It was about threefold faster in cells cultured and tested under 4% O2 as compared to cells cultured and tested under 13% O2, indicating the overall long-term effect. Glycolysis was about twofold faster in cells cultured and tested under 4% O2 as compared to cells cultured under 4% O2 but tested under 13% O2, demonstrating the short-term effect. Glycolysis was about 1.5-fold faster in cells cultured and tested under 4% O2 as compared to cells cultured under 13% O2 but tested under 4% O2, showing the intrinsic long-term effect. This difference was roughly parallel to the difference in levels of glucokinase and pyruvate kinase. Gluconeogenesis, as measured in 'glucagon cells', was stimulated by high O2 levels. Similar to glycolysis overall long-term, short-term and intrinsic long-term effects could be distinguished. The intrinsic long-term effects determined under 13% O2 corresponded to a 1.5-fold stimulation and paralleled the difference in phosphoenolpyruvate carboxykinase levels. The present results show that physiological oxygen concentrations also modulate hepatic carbohydrate metabolism by long-term effects and that the O2 gradient over the liver parenchyma thus contributes to the metabolic differences between periportal and perivenous hepatocytes in vivo.