The beneficial effects of nocturnal oxygen therapy (NOT) in chronic obstructive pulmonary disease (COPD) patients with mild-to-moderate daytime hypoxaemia (arterial oxygen tension (Pa,O2) in the range 7.4-9.2 kPa (56-69 mmHg)) and exhibiting sleep-related oxygen desaturation remains controversial. The effectiveness of NOT in that category of COPD patients was studied. The end points included pulmonary haemodynamic effects after 2 yrs of follow-up, survival and requirement for long-term oxygen therapy (LTOT). Seventy-six patients could be randomized, 41 were allocated to NOT and 35 to no NOT (control). The goal of NOT was to achieve an arterial oxygen saturation of >90% throughout the night. All these patients underwent polysomnography to exclude an associated obstructive sleep apnoea syndrome. The two groups exhibited an identical meansD daytime Pa,O2 of 8.4+/-0.4 kPa (63+/-3 mmHg) at baseline. Twenty-two patients (12 in the NOT group and 10 in the control group, p=0.98) required LTOT during the whole follow-up (35+/-14 months). Sixteen patients died, nine in the NOT group and seven in the control group (p=0.84). Forty-six patients were able to undergo pulmonary haemodynamic re-evaluation after 2 yrs, 24 in the NOT and 22 in the control group. In the control group, mean resting pulmonary artery pressure increased from 19.8+/-5.6 to 20.5+6.5 mmHg, which was not different from the change in mean pulmonary artery pressure in the NOT group, from 18.3+/-4.7 to 19.5+/-5.3 mmHg (p= 0.79). Nocturnal oxygen therapy did not modify the evolution of pulmonary haemodynamics and did not allow delay in the prescription of long-term oxygen therapy. No effect of NOT on survival was observed, but the small number of deaths precluded any firm conclusion. These results suggest that the prescription of nocturnal oxygen therapy in isolation is probably not justified in chronic obstructive pulmonary disease patients.