Low-flow oxygen decreases the frequency of the 3 types of apnea (central, mixed, and obstructive) in patients with predominantly obstructive sleep apnea. The decrease in frequency appears to be accompanied by a shift in apnea distribution, consisting of a decrease in the proportion of central and mixed apneas and an increase in that of obstructive apneas. To determine whether this shift represents a greater inhibitory effect on central and mixed apneas or an increased tendency toward obstructive apneas, we administered low-flow oxygen during sleep to 9 patients who demonstrated predominantly central and mixed sleep apnea (51 +/- 33% and 33 +/- 21% of apneic events, respectively, mean +/- SD) and had resting, room air, oxygen tensions of 83 +/- 11 mmHg. During non-REM sleep, oxygen increased the baseline oxyhemoglobin saturation while reducing the average peak fall in oxyhemoglobin saturation during each apneic event. Oxygen reduced the overall apnea frequency from 66 +/- 7.8 (mean +/- SE) to 43.0 +/- 10.7 episodes per hour (p less than 0.02). Central and mixed apneas decreased markedly from 31.4 +/- 0.6 to 6.4 +/- 4.3 episodes per hour (p less than 0.02) and from 20.9 +/- 5.0 to 4.9 +/- 1.5 episodes per hour (p less than 0.02), respectively. However, obstructive apnea frequency more than doubled from 13.9 +/- 7.0 to 32.1 +/- 9.2 episodes per hour (p less than 0.02). We conclude that in these patients oxygen tension altered both the frequency and distribution of sleep-induced apnea, with a lower oxygen tension increasing the frequency of central and mixed apneas and a higher oxygen tension increasing the frequency of obstructive apneas.(ABSTRACT TRUNCATED AT 250 WORDS)