The efficacy of nasal oxygen during sleep was evaluated in patients with COPD, episodic rapid eye movement sleep desaturation, and a daytime Pa_O2 > 60 mm Hg. The double-blind, randomized 3-yr trial used nasal oxygen versus room air in two groups of nocturnal sleep desaturating subjects. The setting was the outpatient chest clinic of a Veterans Affairs Medical Center. There were 51 patients with moderate to severe COPD, daytime Pa_O2 ⩾ 60 mm Hg: 38 with proven REM sleep desaturation and 13 without desaturation. Nocturnal oxygen at 3 L/min was delivered by concentrator to 19 desaturating subjects, and room air at 3 L/min was delivered by defective concentrator to the remaining 19 desaturating subjects. There was no gas therapy for the 13 nondesaturating subjects. The nocturnal desaturator group who received supplemental oxygen during sleep over 36 months showed a significant downward trend in pulmonary artery pressure (−3.7 mm Hg) compared with desaturating patients treated with room air (+3.9 mm Hg). Nonvascular parameters of hypoxia, such as hemoglobin and red blood cell mass, did not differ between the sham- and oxygen-treated groups. Mortality was decidedly higher in the desaturating patients compared with nondesaturating subjects, but there was no significant difference between oxygen- and sham-treated desaturating subjects. We conclude that nasal supplemental oxygen used during sleep to reverse episodic desaturation in COPD patients whose daytime Pa_O2 is above 60 mm Hg has a beneficial effect in reducing pulmonary artery pressure. Further study is needed to examine the usefulness of supplemental oxygen in reducing mortality In COPD patients with nocturnal desaturation who do not qualify for home oxygen by current criteria.