OBJECTIVE The results of studies on the oxygen response in patients with COPD should provide important clues to the pathophysiology of exertional dyspnoea. We investigated the exercise responses to hyperoxia in relation to dyspnoea profile, as well as cardiopulmonary, acidotic and sympathetic parameters in 35 patients with stable COPD (mean FEV(1) 46% predicted). METHODS This was a single-blind trial, in which patients breathed 24% O(2) or compressed air (CA) in random order during two incremental cycle exercise tests. RESULTS PaO(2) and PaCO(2) were higher (P < 0.0001 and P < 0.05, respectively) at each exercise point while patients were breathing 24% O(2) compared with CA. At a standardized time point near peak exercise, use of O(2) resulted in reduced plasma lactate and plasma noradrenaline concentrations (P < 0.01). Peak minute ventilation/indirect maximum voluntary ventilation was similar while breathing 24% O(2) and CA. At peak exercise, the dyspnoea score, pH and plasma noradrenaline concentrations were similar while breathing 24% O(2) and CA. The dyspnoea-ratio (%) of Δoxygen uptake (peak minus resting oxygen uptake) curve reached a break point that occurred at a similar exercise point while breathing 24% O(2) or CA. CONCLUSIONS Regardless of whether they breathed CA or 24% O(2) , patients with COPD did not develop ventilatory compensation for exertional acidosis, and the pH values measured were similar. Hyperoxia during a standardized exercise protocol did not alter the pattern of exertional dyspnoea in these patients, compared with breathing CA, although hyperoxia resulted in miscellaneous effects.