The aim of this study was (1) to evaluate the effects of long-term oxygen treatment (LTOT) in elderly patients with severe exacerbations of chronic obstructive pulmonary disease (COPD) and hypoxaemia, (2) to study the health-related quality of life (QOL) during hospital stay and at follow-up, (3) to study the safety of an oxygen withdrawal test performed a few days after admission to hospital and the possibility to predict the future need for LTOT from that test. Patients > 70 years with COPD-exacerbations with hypoxaemia were included after 5-7 days treatment in hospital. Inclusion was based on results of a standardised oxygen withdrawal test. After 1, 3, 6 and 12 months a new oxygen withdrawal test was performed. Health-related QOL was evaluated with SF-36 and the St George's Respiratory Questionnaire at inclusion and after 3,6 and 12 months. Forty-seven patients were screened for participation and 29 patients, mean age 79 years, participated in the study. Only one patient could not tolerate the oxygen withdrawal test. Eighteen patients survived to the follow-up after 12 months, 8/19 women and 2/10 men died. After 1 month LTOTwas needed (PaO2 without oxygen was < or = 75 k Pa) in only 6/20 patients. The effect of LTOT could therefore not be studied. Most components of SF-36 were very low at inclusion, but tended to increase after 3 months and were among the surviving patients after 12 months similar to that of healthy people of the same age for psychic well being and functioning. Especially, the symptom score of the SGRQ improved after 3 months. In conclusion, the future need for LTOT cannot be judged after a few days treatment in hospital due to exacerbations with hypoxaemia in elderly patients with COPD. A standardised oxygen withdrawal test can be safely done. Health-related QOL is low in patients during the stay in hospital, but improves after returning home.