Pulse delivery (PD) of oxygen was compared with continuous flow (CF) utilizing transtracheal oxygen catheter (TTO) and nasal cannula (NC) in 20 stable patients with chronic hypoxemia. Oxygen saturation, respiratory rate, and accuracy of pulsed oxygen delivery were measured during sleep studies and these parameters, as well as arterial blood gases, were evaluated during rest and exercise. Additionally, bulk liquid oxygen use was measured under each condition, for a period of 1 month. Pulse delivery NC was evaluated in six subjects, CF NC in 14 subjects, and PD and CF TTO in 20 subjects over the 1-month period. Results showed that, as a group, patients were adequately oxygenated when utilizing the PD with both NC and TTO as assessed by arterial blood gases, oximetry, and hematocrit. However, four subjects could not be adequately oxygenated on PD NC during exercise even at the maximum liter per minute setting and could not be studied with this mode of therapy. The PD settings in the remaining subjects were equivalent to continuous flow settings for TTO and NC as assessed by PaO2 for rest and SaO2 for exercise and sleep. Compared with standard CF NC, the daily bulk oxygen use was decreased by 29.4 percent with CF TTO, by 48.2 percent with PD NC, and by 49.9 percent with PD TTO. We conclude that, compared with CF NC, PD of oxygen via TTO or NC by this method appears to be a safe, reliable, effective, and cost-effective method of oxygen delivery in the majority of subjects when used with proper screening.