BACKGROUND: A retrospective analysis was conducted to determine whether oxyhemoglobin saturation measured by pulse-oximetry (SpO2) during a 5-minute corridor walk (WALK5:00) accurately predicts oxygen flow required by hypoxemic patients with chronic obstructive pulmonary disease (COPD) using the standard nasal cannula (SNC) and reservoir nasal cannula (RNC) to maintain average SpO2 ≥ 91% during simulated activities of daily living (ADLs). METHODS: Oxygen dose was titrated in 15 men who were hypoxemic (average SpO2 ≤ 88%) at rest or with mild exercise, mean (SD) age 69 (9) years, with 40 (17) pack-year smoking history, FEV1/FVC% 41 (10)%, PaO2 57 (4) torr, and exertional dyspnea. Each subject used both devices, on separate occasions, in a randomized order, during incremental cycle ergometry and a 4-station circuit simulating ADLs. RESULTS: Average SpO2 during the last 30 seconds of WALK5:00 (WALK:30) was significantly greater than during the lift-push-carry, undress and dress, and ascending and descending stairs. No significant difference between average SpO2 for WALK:30 [SNC = 89 (3)%, RNC = 90 (3)%] and bicycle ergometry [SNC = 91 (3)%, RNC = 92 (3)%] was found. CONCLUSION: SNC and RNC oxygen flow prescriptions intended to maintain SpO2 ≥ 91% during physical exertion in hypoxemic COPD patients may be determined during a 5-minute corridor walk using only the average of SpO2S taken during the last 30 seconds of exercise. [Respir Care 1996; 41(1):30-36]