The effect of sepsis on tissue oxygenation is complex and often confusing. This is in part due to the paucity of data from patients (excluding whole-body studies), and in part to the huge variation in laboratory models which have utilised different species, organs and cell lines, different insults, study durations and methods, fluid resuscitation regimens, and end-points. An organ-variable redistribution in microvascular flow undoubtedly occurs though its significance in causing tissue hypoxia remains questionable. A lower oxygen consumption is recognised in patients with an increasing severity of sepsis, with a better prognosis being seen in those spontaneously achieving higher values of global oxygen delivery and consumption. However, septic patients and (well-resuscitated) animal models demonstrate an increase in tissue oxygen tension. This suggest adequate delivery but impaired utilisation of oxygen. As over 90% of oxygen consumed by the body is for mitochondrial generation of ATP by oxidative phosphorylation, a cellular defect is strongly implicated. Mitochondrial dysfunction has been inconsistently demonstrated; on closer inspection, short-term and/or ‘milder’ insult models often show no change or an increase in mitochondrial activity whereas the longer duration and/or more severe models show depressed function. This is borne out by preliminary human studies. Reactive oxygen and nitrogen species appear instrumental in causing this dysfunction; they may also act as signalling mediators for cellular ‘shutdown’, possibly as a protective mechanism to reduce ATP turnover in the face of a decrease in production.