The number of deaths from carbon monoxide (CO) poisoning remained constant for many years but has recently started to fall. There are three reasons. Firstly, it has become more fashionable to commit suicide by more sophisticated methods, particularly now that tranquillisers and antidepressants are widely prescribed. Secondly, Britain's domestic gas supply is being changed over from coal gas to natural gas, which contains no carbon monoxide-though so far only an eighth of the country has actually been converted to the new gas. Thirdly, and perhaps most significant, is the changeover which has taken place in the manufacture of gas for domestic use, coal carbonization having given way to oil reforming. Techniques for reducing the amount of CO present have also been introduced, but as G. Thurston' has shown, inadequate combustion or ventilation can be lethal even when the original fuel contains no CO. So in about eight years, when the whole country has been converted to natural gas, there will still be numbers of patients with CO poisoning needing resuscitation.Recently J. S. Smith and S. Brandon2 reviewed 206 episodes of CO poisoning in Newcastle. The mortality rate was 3900. In 20% of patients recovery was complicated by prolonged delirium suggesting that "all degrees of functional or structural neurological damage may have occurred, yet oxygen therapy was given in only 43.8% of suicidal and 32-5% of accidental exposures." They suggest that in view of the risk of persistent neuropsychiatric sequelae current patterns of management should be revised.In selecting a form of therapy there are two aspects to consider: firstly, the prevention of death and, secondly, the reduction of neuropsychiatric sequelae such as those described by H. Garland and J. Pearce.3 Carboxyhaemoglobin should be eliminated as quickly as possible because its presence alters the dissociation curve of the remaining oxyhaemoglobin, impeding oxygen release to the tissues.4 J. S. Haldane5 was the first to describe the use of hyperbaric oxygen in CO poisoning, when he failed to poison a mouse placed in a jar containing two atmospheres of oxygen and one of CO. This was all the more remarkable for the fact that the affinity of haemoglobin for CO is about 250 times that for oxygen. Hyperbaric oxygen keeps the patient's tissues oxygenated by the oxygen physically dissolved in the plasma at a time when his haemoglobin is not available to him for oxygen transport. It is also the most efficient means of reducing the carboxyhaemoglobin level, as was shown by T. A. Douglas and his colleagues6 in 1962 when they compared the efficiencies of oxygen at one atmosphere's pressure, oxygen at two atmospheres' pressure, and of 50/% and 70 % carbon dioxide in oxygen.