OBJECTIVE To assess cardiorespiratory, tissue oxygen and hepatic nicotine adenine dinucleotide hydride (NADH) responses to graded hypoxia. METHODS Prospective, controlled, randomized study. METHODS University laboratory. METHODS 18 anaesthetised Sprague-Dawley rats spontaneously breathing either 21% (controls), 12.5% or 10% inspired oxygen concentrations (6 rats per group). RESULTS All animals in the 21 and 12.5% O2 groups survived the 3-h study period, compared to only 1 in the 10% O2 group. In this latter group, mean arterial pressure and renal blood flow fell immediately with hypoxaemia, whereas aortic blood flow was maintained until the preterminal stages. Critical cellular hypoxia was suggested by an increasingly severe base deficit, an initial rise then a preterminal fall in hepatic NADH intensity and premature death in all but 1 animal. Hepatic NADH fluorescence intensity was unchanged in control animals but showed a progressive rise in the 12.5% O2 group, accompanied by a small though static increase in arterial base deficit. No significant differences were seen in arterial and tissue partial pressure of oxygen between the 12.5 and 10% O2 groups. CONCLUSIONS This study demonstrates major differences in cardiorespiratory, hepatic NADH and outcome responses to small variations in the degree of hypoxic hypoxia. The fall in NADH fluorescence intensity presages impending death and is likely to reflect failure of cellular metabolic processes.