Criteria for the prescription of long-term oxygen therapy (LTOT) have been published by academic societies and regulatory bodies, but many prescriptions for LTOT do not fulfil these criteria. Demographic, functional data and survival were compared in chronic obstructive pulmonary disease (COPD) patients with different levels of oxygenation, i.e. arterial oxygen tension (Pa,O2) < 8 kPa or > or = 8 kPa (60 mmHg), at the time of initial registration in the ANTADIR Observatory. Data were collected between 1984-1995. Selection criteria were a diagnosis of COPD or emphysema with forced expiratory volume in one second (FEV1) < 80% pred, FEV1/vital capacity (VC) < 70% and age between 18-75 yrs. Of 7,700 patients prescribed LTOT 18.5% had stable Pa,O2 > or = 8 kPa. While the FEV1 was the same they differed from the patients with more severe hypoxaemia in having a higher rate of diagnosis of primary emphysema and a lower arterial carbon dioxide tension (Pa,CO2). In this group of patients LTOT was more frequently administered as liquid oxygen than in other patients on LTOT. The survival of these patients was reduced compared to the general population of the same age and sex but comparable to that of patients with a Pa,O2 between 6.7-8 kPa (50-60 mmHg). Patients prescribed long-term oxygen therapy with an arterial oxygen tension > or = 8 kPa (60 mmHg) in the ANTADIR network were shown to have severe chronic obstructive pulmonary disease on the basis of spirometry and their survival was similar to that of more hypoxaemic patients. Randomized controlled trials of the effect of long-term oxygen therapy in patients with arterial oxygen tension > or = 8 kPa are needed.