It has been hypothesized but not firmly established that sleep-related hypoxaemia could favour the development of pulmonary hypertension in chronic obstructive pulmonary disease (COPD) patients without marked daytime hypoxaemia. We have investigated the relationships between pulmonary function data, sleep-related desaturation and daytime pulmonary haemodynamics in a group of 94 COPD patients not qualifying for conventional O2 therapy (daytime arterial oxygen tension (Pa,O2) in the range 7.4-9.2 kPa (56-69 mmHg)). Nocturnal desaturation was defined by spending > or = 30% of the recording time with a transcutaneous O2 saturation < 90%. An obstructive sleep apnoea syndrome was excluded by polysomnography. Sixty six patients were desaturators (Group 1) and 28 were nondesaturators (Group 2). There was no significant difference between Groups 1 and 2 with regard to pulmonary volumes and Pa,O2 (8.4+/-0.6 vs 8.4+/-0.4 kPa (63+/-4 vs 63+/-3 mmHg)) but arterial carbon dioxide tension (Pa,CO2) was higher in Group 1 (6.0+/-0.7 vs 53+/-0.5 kPa (45+/-5 vs 40+/-4 mmHg); p<0.0001). Mean pulmonary artery pressure (Ppa) was very similar in the two groups (2.6+/-0.7 vs 2.5+/-0.6 kPa (19+/-5 vs 19+/-4 mmHg)). No individual variable or combination of variables could predict the presence of pulmonary hypertension. It is concluded that in these patients with chronic obstructive pulmonary disease with modest daytime hypoxaemia, functional and gasometric variables (with the noticeable exception of arterial carbon dioxide tension) cannot predict the presence of nocturnal desaturation; and that mean pulmonary artery pressure is not correlated with the degree and duration of nocturnal hypoxaemia. These results do not support the hypothesis that sleep-related hypoxaemia favours the development of pulmonary hypertension.