In patients with chronic obstructive pulmonary disease (COPD), exercise-induced hypoxemia (EIH) is an important limiting factor of exercise performance, but there are very few reports of the prognostic value of EIH. We therefore examined whether EIH is related to long-term outcome in patients with COPD. We gathered data on survival of 77 patients with COPD who had undergone three-minute incremental treadmill exercise tests between 1979 and 1988. A plastic catheter was placed percutaneously into a brachial artery to measure arterial blood gases at each stage of exercise. Expired gas was analyzed continuously during exercise with a Respiromonitor RM-300 (Minato Medical Science). As an index of the severity of EIH we used the delta PaO2/delta VO2 (mmHg/L/min), PaO2-slope. Patients were assigned to two groups according to PaO2-slope, and actuarial survival curves were plotted for the two groups. The survival data were analyzed by generalized Wilcoxon analysis. Survival was significantly better in the group with low PaO2-slopes (< 20 mmHg/L/min) than in the group with high PaO2-slopes (> or = 20 mmHg/ L/min) (p = 0.0031). Also, among the patients in whom PaO2 during exercise was less than 60 mmHg, there was a significant difference in survival between those who received home oxygen therapy and those who did not. We conclude that the degree of EIH can be used to predict survival in patients with COPD, and that the degree of EIH should be considered when prescribing continuous home oxygen therapy for these patients.