It is unclear whether short-term benefits from supplemental oxygen translate into improved quality of life in patients with severe COPD. In a 12 wk double-blind randomized crossover study, we assessed the effects of supplemental air and oxygen on exercise performance (step tests and 6 min walking distance 6MWD) initially and after two 6 wk periods at home using exertional cylinder air or oxygen. We measured quality of life at baseline and after the two 6 wk domiciliary periods. The 26 patients (24 males) had a mean age of 73 +/- 6 yr; mean FEV1, 0.9 +/- 0.4 L; mean DLCO, 10.6 +/- 2.4 ml/min/mm Hg; mean resting PO2, 69 +/- 8.5 (range 58 to 82) mm Hg; mean PCO2, 41 +/- 3.3 mm Hg; and mean resting SaO2, 94 +/- 2.1 (mean +/- SD). Laboratory tests were performed breathing intranasal air or oxygen at 4 L/min, and measurements were made of SaO2 and Borg dysnea scores. Supplemental oxygen increased 6MWD and steps by small, statistically significant increments acutely at baseline and after 6 and 12 wk, without corresponding falls in Borg score. Degree of desaturation at baseline did not correlate with increase in 6MWD or steps achieved at baseline or at 6 or 12 wk, nor with the domiciliary gas used. There was no difference in 6MWD or steps achieved while breathing supplemental oxygen after 6 wk of domiciliary oxygen compared with domiciliary air. Small improvements in quality of life indices were found after domiciliary oxygen, and mastery also improved after domiciliary air. There were no differences in quality of life, however, when domiciliary oxygen was compared with domiciliary air. Although oxygen supplementation induced small acute increments in laboratory exercise performance, such improvements had little impact on the patients' daily lives.