1. The hypermetabolism frequently observed at rest in patients with chronic obstructive pulmonary disease has been attributed to a high cost of breathing. However, measurement of the cost of breathing by the usual hyperventilation procedure is fraught with methodological problems. The purpose of this study was to measure more directly the cost of breathing in a group of ambulatory patients with stable chronic obstructive pulmonary disease. 2. The cost of breathing was calculated as the difference in oxygen consumption measured by indirect calorimetry between spontaneous breathing and noninvasive mechanical ventilation. Inspiratory muscle rest was achieved by negative or positive pressure ventilation and assessed by the recording of surface electromyograms of the diaphragm and parasternal intercostal muscles. 3. Seven tests were performed in six ambulatory patients with stable chronic obstructive pulmonary disease, four tests using positive pressure ventilation and three with negative pressure ventilation. During mechanical ventilation, the electromyographic activity of the diaphragm decreased by 70 +/- 22%, while that of the parasternals was suppressed in four tests, and remained unchanged in three. However, oxygen consumption was only 1.6 +/- 6.2% lower during mechanical ventilation. 4. The cost of breathing measured in this study was therefore much lower than previously published values. Stress was not likely to influence the results, as both the heart rate and plasma catecholamines did not change between spontaneous breathing and mechanical ventilation. These results suggest that the cost of breathing in ambulatory patients with stable chronic obstructive pulmonary disease may be lower than previously estimated.