The methylxanthine derivatives are known to have respiratory stimulant properties. To determine whether these drugs would improve obstructive sleep apnea, 10 male patients with obstructive sleep apnea (OSA) (Apnea Index > 15/h) were given infusions of aminophylline and a saline placebo on 2 separate nights a week apart, using a randomized crossover design. There was a significant decrease during aminophylline infusion in the frequency of those apneas, which contained periods of complete respiratory inactivity (central and mixed apneas; placebo, 4.3 ± 1.8/h; aminophylline, 0.7 ± 0.5/h; p < 0.05). There was no change in either the frequency (placebo, 31.8 ± 5.9/h; aminophylline, 28.7 ± 8.7/h; NS) or duration of obstructive apneas. Mean and minimal arterial oxygen saturation values were also unchanged. Sleep architecture was markedly disturbed by aminophylline. There was a reduction in sleep efficiency (placebo, 84.8 ± 2.0%; aminophylline, 60.2 ± 5.0%; p < 0.005), an increase in sleep fragmentation (sleep stage shifts/h: placebo, 11.6 ± 1.3; aminophylline, 21.0 ± 2.9; p < 0.05) and less Stage 2 and more Stage 1 non-REM sleep. We conclude that aminophylline reduces central apnea and the central component of mixed apneas but has no effect on obstructive apnea. Theophylline is therefore unlikely to be therapeutically useful in patients with OSA, and because it leads to marked sleep disruption, its long-term use could conceivably increase the propensity to upper airway occlusion during sleep.