Five different oxygen-conserving devices were tested in each of ten oxygen-dependent patients with COPD who had met the NOTT criteria for continuous oxygen use. They were tested on room air, their prescribed continuous oxygen flow and then on each of the five devices. The devices which delivered a bolus of oxygen during early inspiration or increased oxygen delivery as the respiratory rate increased did better than those devices which delivered oxygen at a normal flow rate during inspiration or a fixed portion of inspiration. In at least one of the subjects each device was associated with desaturation to less than 80 percent during a 12-min walk. It is concluded that oxygen-conserving devices vary in their ability to maintain SaO2 levels during exercise. It is recommended that a home oxygen evaluation include measurement of an exercise SaO2 utilizing the prescribed oxygen delivery system.