Résumé : The effects of exercise training were investigated in 11 male patients with chronic lung diseases, and the results were compared with those obtained from 5 untrained control patients. Using symptom limited oxygen consumption as a criterion, the average improvement was 10% in the first group and 0% in the control group. Of the 11 trained patients, 4 showed an average improvement of 20.5% based entirely on an increase in the maximal arteriovenous oxygen difference. Maximal cardiac output and stroke volume did not change. The degree of improvement was related to the arterial blood P(O 2) during exercise and P(CO 2) at rest measured before training. After training, a favorable evolution of the arterial blood P(CO 2) at rest and of the mean arterial pulmonary pressure was observed. This increased P(O 2) at rest after training suggested an improvement of the ventilation/perfusion ratio. No effect was observed on the pulmonary volumes and on the pulmonary diffusing capacity after training. The modifications observed after training were not related to the natural evolution of the disease, as demonstrated by the constant results observed before and after the control period in the untrained patients. The principal effects of training in the pulmonary patients were an increased ability for the muscles to extract oxygen more thoroughly from the circulating blood and a favorable evolution of the ventilation/perfusion ratio. These effects are, nevertheless, variable from one patient to another. (38 references.)