The follow-up of 87 patients with long-term oxygen therapy over a 3-year period revealed a mortality of 23.7% during the first year, of 37.4% after 2 years and of 45% after 3 years, respectively. The retrospective analysis of baseline characteristics in order to evaluate the causes for early deaths showed no difference in age, lung function tests nor the degree of respiratory insufficiency documented by serial blood gas analysis between the survivors (n = 61) and the dead (n = 26). However, in the group of early deaths there was a significant increase in pulmonary hypertension (mean pulmonary arterial pressure 40.1 +/- 12.0 mm Hg) compared with the survivors, who had only mild hypertension (mean pulmonary arterial pressure 29.4 +/- 8.9 mm Hg). In addition there was also a lesser response in pulmonary vasodilation during oxygen breathing in the patients who died within the first 2-year period. We conclude that marked pulmonary hypertension with poor response to oxygen breathing is a major limiting factor in the prognosis of patients with long-term oxygen therapy. In contrast neither severe airway obstruction nor pronounced hypoxemia are related to early mortality and should therefore not be considered as a criterion for rapid deterioration and fatal outcome despite oxygen therapy.