The effect of nocturnal oxygen therapy on mortality and morbidity rates and on progression of disease was tested in a double-blind, randomized trial of 28 subjects with advanced cystic fibrosis. Patients were selected on the basis that their awake arterial oxygen tension was less than 65 mm Hg when they were clinically stable. Oxygen was prescribed in 1 L/min increments to obtain an awake arterial oxygen tension of greater than or equal to 70 mm Hg. Subjects received humidified oxygen or room air from modified concentrators. They were enrolled over a 3-year period and followed for an average (+/- SD) of 26 +/- 9 months. The average number of hours per night of concentrator use was 5.3 +/- 3.2 hours in the air group and 7.0 +/- 1.9 hours in the oxygen group. Over the follow-up period there were four deaths in each group, and oxygen therapy had no significant effect on the frequency of hospitalizations. Progression of disease was ascertained from nutritional status, pulmonary function, blood gas values, exercise ability, and right ventricular ejection fraction response to exercise (as measured by equilibrium-gated radionuclide angiocardiography), and psychologic status was measured by standardized tests of mood, self-esteem, and cognitive function; group comparisons for the first year revealed no significant differences; however, school or work attendance was maintained in the oxygen group but deteriorated in the air group. Clinical signs of cor pulmonale were documented during follow-up in 10 patients in toto, and all lived at least 9 months from the onset of these signs. The lack of association between the onset of these signs and imminent death, or the usefulness of measurements of the maximal oxygen uptake during progressive exercise and the right ventricular ejection fraction response to exercise as prognostic indicators, suggest that death may not be the result of cor pulmonale. We conclude that nocturnal oxygen treatment in patients with cystic fibrosis did not appear to affect mortality rates, the frequency of hospitalizations, or the progression of disease; oxygen use should be instituted only after the development of symptoms related to hypoxemia.