The effect of a 24-h controlled-release (CR) preparation of theophylline (Th) on nocturnal arterial oxygen saturation (SaO2) and sleep quality has been evaluated in 7 patients with advanced chronic obstructive pulmonary disease (COPD) in a double-blind cross-over experiment; median values: age 61 y; PaO2 8.0 kPa; PaCO2 5.8 kPa. During treatment with 450-900 mg Th in the evening, morning plasma drug levels ranged from 5.2-12.9 micrograms.ml-1. During Th and placebo treatment, the median evening FEV1 was 0.45 l and 0.46 l, respectively, and the morning FEV1 was 0.53 l and 0.41 l. Sleep was monitored by whole-night polysomnographic recording of oximetric SaO2, airflow, respiratory and body movements (static charge sensitive bed), eye movements and submental electromyogram. There was no significant difference between Th and placebo in sleep quality. Th treatment was associated with a marginal improvement in nocturnal oxygenation in most of the patients; the average nocturnal SaO2 ranged from 84.4%-92.8% during Th and from 82.2-90.5% during placebo treatment, respectively. Only in the morning, during the last 2 h in bed, was the slight difference in SaO2 statistically significant in favour of Th. It is concluded that a moderate dose of CR-theophylline did not alter the sleep quality or substantially improve nocturnal oxygenation in patients with advanced COPD and mild to moderate day-time hypoxaemia.