We thank Drs Patil and Maloney for their comment that supplementary oxygen may be beneficial for up to ten minutes after nitrous oxide anaesthesia and we would hope that continuous monitoring of oxygen saturation postoperatively will ensure that the oxygen dose is titrated to the patient's needs in these particular circumstances. Dr Gwinnutt suggests that our article 1 over-emphasised the risks of hyperoxaemia and type-2 respiratory failure but under-estimated the risks of hypoxaemia amongst patients cared for by anaesthetic colleagues. We agree that the best known example of oxygen sensitivity is the patient with acute exacerbation of chronic obstructive pulmonary disease (COPD), a condition much less common in anaesthetic and surgical practice than on medical wards. However, anaesthetists need to be aware that approximately 9% of UK adults aged > 65 have evidence of severe or very severe COPD on routine spirometry and about 2% of UK adults have morbid obesity, another condition where supplementary oxygen may cause hypercapnic respiratory failure 2-4. Patients with severe skeletal deformity or neuromuscular conditions causing hypercapnic respiratory failure may also present for surgery. In addition to these at-risk groups, even healthy adults can develop hypercapnic respiratory failure due to hypoventilation if given opioids. Audits at our institution found that 13% of blood gas specimens from surgical wards were hypercapnic (PCO2 > 6.0 kPa) compared with 29% of specimens from patients on medical wards, confirming that type-2 respiratory failure is less common (but not rare) on surgical wards. Type-2 respiratory failure was twice as common as type-1 (PO2 < 8.0 kPa with normal or low PCO2) on a hospital-wide basis 5, 6. We have analysed this audit database further and the results are shown in Table 1: type-2 respiratory failure is actually more common than pure hypoxaemia (type-1 respiratory failure) in blood gas samples from surgical wards, surgical HDU, theatres and ICU. Type-2 respiratory failure complicated by respiratory acidosis was commoner than type-1 failure on surgical HDU and in theatre and ICU and many additional patients had a mixed metabolic and respiratory acidosis. Our data also demonstrated that many patients in surgical and anaesthetic areas were hyperoxaemic including a high proportion with type-2 respiratory failiure. A random sample of casenotes identified risk factors for type-2 respiratory failure in 18% of the patients who required blood gas analysis on surgical wards and 23% from theatre samples. Education of medical students and junior doctors will need to reflect that type-2 respiratory failure is commoner than type-1, and that many patients with type-2 respiratory failure are hyperoxaemic due to oxygen therapy. Gwinnutt has not presented any evidence of benefit from the routine use of supplementary oxygen in postoperative care as distinct from our proposal to monitor patients carefully and give this drug to those who need it. The lack of likely benefit of routine supplemental postoperative oxygen, together with the potential hazards, were reviewed in detail by Downs and Rothen, who are both trained in anaesthesia 7, 8. We agree strongly with Gwinnutt that pulse oximetry and the administration of supplemental oxygen is only a small part of the assessment and resuscitation of at-risk patients, as discussed in the second paragraph of our editorial. We would guide readers to section 7 of the British Thoracic Society Emergency Oxygen Guideline which is devoted to the detailed clinical and laboratory assessment of seriously ill patients, and to Box 4 in our recent review of emergency oxygen use where we discuss several ways to increase tissue oxygen delivery, including the importance of airway management and blood transfusion when appropriate 9, 10.