STUDY OBJECTIVES: To assess oxygen desaturation during activities and to evaluate the short-term effects of supplemental O(2) use in patients with severe COPD who do not qualify for long-term O(2) therapy. DESIGN: A double-blind, randomized, placebo-controlled trial. SETTING: Outpatients from the pulmonary diseases division of a tertiary-care university hospital. PATIENTS: Twenty patients with stable COPD with FEV(1)/FVC ratios of < 50%, FEV(1) levels < 55% of the predicted normal value, and PaO(2) levels of > 60 mm Hg when resting. INTERVENTIONS: Patients were initially evaluated with pulmonary function tests, blood gas analysis, and Doppler echocardiography, and they underwent the following three 6-min walking tests (WTs) in a random sequence: basal WT (BWT); WT while breathing compressed air (CAWT); and WT while breathing O(2) (O(2)WT). MEASUREMENTS AND RESULTS: The distance walked was recorded in meters. Dyspnea was measured by Borg scale measurement before and after the tests, and arterial oxygen saturation measured by pulse oximetry (SpO(2)) was continuously monitored. Results were analyzed by grouping patients in the following manner: desaturators (DSs) (ie, patients with a drop in SpO(2) of at least 5% and < 90% during the WT) vs nondesaturators (NDSs); and O(2) responders (ie, patients with an increase of at least 10% in the distance walked and/or a decrease of at least 3 points in Borg index score) vs nonresponders. During the BWT, 11 of 20 patients (55%) were defined as desaturators. During the O(2)WT, the SpO(2) remained at > 90% in every patient. The distance walked increased by 22% (p < 0.02), and dyspnea decreased 36% (p < 0.01) in DS patients. In NDS patients, O(2) administration reduced dyspnea by 47% (p < 0.001), but the distance walked did not improve. Responses were markedly different from one patient to another. No significant differences were noticed between the results of the BWT and CAWT in any of the groups. Thirteen O(2) responders did not differ from 7 nonresponders either in basal data or in desaturation measure during the BWT, except that all walking responders (five patients) were above the median of basal left ventricle performance. CONCLUSIONS: Most of the studied COPD patients desaturated during the BWT. O(2) administration avoided desaturation and could increase the distance walked and reduce dyspnea, but these effects were not related to walking desaturation in individual cases. Improvements were not a placebo effect. The therapeutic role of O(2) during activities in some patients with severe COPD needs to be individually assessed.