Dysoxia can be defined as ATP flux decreasing in proportion to O₂ availability with preserved ATP demand. Hepatic venous -hydroxybutyrate-to-acetoacetate ratio (-OHB/AcAc) estimates liver mitochondrial NADH/NAD and may detect the onset of dysoxia. During partial dysoxia (as opposed to anoxia), however, flow may be adequate in some liver regions, diluting effluent from dysoxic regions, thereby rendering venous -OHB/AcAc unreliable. To address this concern, we estimated tissue ATP while gradually reducing liver blood flow of swine to zero in a nuclear magnetic resonance spectrometer. ATP flux decreasing with O₂ availability was taken as O₂ uptake (O₂) decreasing in proportion to O₂ delivery (O₂); and preserved ATP demand was taken as increasing P_i/ATP. O₂, tissue P_i/ATP, and venous -OHB/AcAc were plotted against O₂ to identify critical inflection points. Tissue dysoxia required mean O₂ for the group to be critical for both O₂ and for P_i/ATP. Critical O₂ values for O₂ and P_i/ATP of 4.07 ± 1.07 and 2.39 ± 1.18 (SE) ml · 100 g¹ · min¹, respectively, were not statistically significantly different but not clearly the same, suggesting the possibility that dysoxia might have commenced after O₂ began decreasing, i.e., that there could have been "O₂ conformity." Critical O₂ for venous -OHB/AcAc was 2.44 ± 0.46 ml · 100 g¹ · min¹ (P = NS), nearly the same as that for P_i/ATP, supporting venous -OHB/AcAc as a detector of dysoxia. All issues considered, tissue mitochondrial redox state seems to be an appropriate detector of dysoxia in liver.