‘Oxygenation’ failure is both a common cause and a consequence of critical illness, and the optimisation of oxygen delivery is a key and central tenet of intensive care medicine. However, the best strategy for achieving this goal has proved remarkably difficult to define. Transfusion and supra-normalisation of cardiac index have both been associated with unexpected harm in some categories of critically ill patient.1–3 However, perhaps the most fundamental weakness in the supportive care evidence base is our inability to recommend an optimal PaO₂/FiO₂ ratio to best balance the risks and benefits of supplemental oxygen therapy. This is because too much and too little oxygen can both result in host injury in the setting of critical illness. Efforts to avoid injurious ventilator settings, unopposed atelectasis, vasoconstriction, loss of hypoxic ventilatory drive and free radical injury are offset by the known benefits of enhanced oxygen delivery and the discovery that ‘permissive’ hypoxia may have long term neurocognitive sequelae.4–7

The presence of traumatic brain injury (TBI) raises the stakes and the complexity of this risk-benefit equation. It is unsurprising that evidence surrounding optimal oxygenation of brain-injured patients is inconsistent. While some individual institutions have reported a benefit from brain …