Administration of nocturnal oxygen for 1 night to patients with obstructive sleep apnea (OSA) causes a moderate reduction in apnea frequency without improving hypersomnolence. Therefore, we administered oxygen chronically to patients with OSA to determine: whether apnea frequency would be further reduced, whether the effect of oxygen upon apnea frequency is correlated with an increased ventilatory response to hypoxia and hypercapnia, and whether hypersomnolence improves with more prolonged oxygen administration. In a single-blinded, nonrandomized trial, we compared the effects of 1 month of oxygen (4 L/min by nasal cannula) with room air (4 L/min by nasal cannula) placebo during sleep in 7 men and 1 woman with obstructive sleep apnea. During non-REM sleep, acute oxygen administration elevated the average low oxy-hemoglobin saturation during apneic events and decreased apnea frequency. These acute effects persisted during chronic oxygen administration but reverted to the preoxygen effects immediately upon discontinuing oxygen. One month of oxygen did not affect the waking ventilatory response to hypoxia or hypercapnia; however, waking PaCO2 increased from 40 +/- 1 mm Hg (mean +/- SE) after placebo to 43 +/- 1 mm Hg after oxygen (p less than 0.01). Neither subjective nor objective hypersomnolence consistently improved after 1 month of oxygen administration. We conclude that: first, oxygen has no effect upon apnea frequency beyond the period of administration, and the reduction of apnea frequency is not correlated with an increased sensitivity to chemical ventilatory stimuli. The reduced apnea frequency may be related to an increased PaCO2 stimulating ventilation during sleep.(ABSTRACT TRUNCATED AT 250 WORDS)