A controlled trial of long term domiciliary oxygen therapy has been carried out in three centres in the U.K. The 87 patients, all under 70 years of age, who took part had chronic bronchitis or emphysema with irreversible airways obstruction, severe arterial hypoxaemia, carbon dioxide retention, and a history of congestive heart failure. The patients were randomised to oxygen therapy (treated) or no oxygen (controls). Oxygen was given by nasal prongs for at least 15 h daily, usually at 2 1/min. The two groups were well matched, both clinically and in terms of lung function and other laboratory findings. 19 of the 42 oxygen treated patients died in the five years of survival follow-up compared with 30 out of 45 controls: in the 66 men in this trial the survival advantage of oxygen did not emerge until 500 days had elapsed. Survival for the 12 female controls was surprisingly poor, 8 of them being dead at 3 years. Mortality was not easy to predict, though a summation of arterial carbon dioxide tension and red cell mass was helpful. Neither time spent in hospital because of exacerbations of respiratory failure nor work attendance were affected by oxygen therapy, but these patients were very ill at the start of the trial and many had already retired on grounds of age or ill-health. Physiological measurements suggested that oxygen did not slow the progress of respiratory failure in those who died early. However, in longer term survivors on oxygen, arterial oxygenation did seem to stop deterioration.