This study aimed to review the evidence for the use of long-term oxygen therapy for patients with chronic obstructive pulmonary disease (COPD). The design was a systematic Cochrane review of randomized controlled trials (RCTs) of long-term oxygen therapy for COPD and main outcome measure was survival on home oxygen therapy. Five RCTs were identified. Data from two trials of nocturnal oxygen therapy in mild to moderate hypoxaemia were aggregated. Data from the other three trials could not be aggregated because of differences in trial design and patient selection. Treatment with continuous versus nocturnal oxygen therapy produced a significant improvement in mortality after 24 months [Peto odds ratio 0.45, 95% confidence interval (95% CI) 0.25-0.81] for the continuous therapy group. Treatment with oxygen therapy versus no oxygen therapy showed a significant improvement in mortality after five years in the group receiving oxygen therapy (Peto odds ratio 0.42, 95% CI 0.18-0.98). There was no difference in mortality for patients with COPD and mild to moderate daytime hypoxaemia and nocturnal desaturation receiving nocturnal oxygen therapy versus no oxygen therapy or sham treatment. Long-term oxygen therapy versus no oxygen therapy in patients with COPD and moderate hypoxaemia had no effect on survival. In conclusion, long-term oxygen therapy improved survival in a selected group of COPD patients with severe hypoxaemia but few co-morbidities. Long-term oxygen therapy did not improve survival in patients with moderate hypoxaemia or in those with mild to moderate hypoxaemia and arterial desaturation at night.