Anecdotal observations suggest that hypoxia does not elicit dyspnea. An opposing view is that any stimulus to medullary respiratory centers generates dyspnea via "corollary discharge" to higher centers; absence of dyspnea during low inspired Po(2) may result from increased ventilation and hypocapnia. We hypothesized that, with fixed ventilation, hypoxia and hypercapnia generate equal dyspnea when matched by ventilatory drive. Steady-state levels of hypoxic normocapnia (end-tidal Po(2) = 60-40 Torr) and hypercapnic hyperoxia (end-tidal Pco(2) = 40-50 Torr) were induced in naive subjects when they were free breathing and during fixed mechanical ventilation. In a separate experiment, normocapnic hypoxia and normoxic hypercapnia, "matched" by ventilation in free-breathing trials, were presented to experienced subjects breathing with constrained rate and tidal volume. "Air hunger" was rated every 30 s on a visual analog scale. Air hunger-Pet(O(2)) curves rose sharply at Pet(O(2)) <50 Torr. Air hunger was not different between matched stimuli (P > 0.05). Hypercapnia had unpleasant nonrespiratory effects but was otherwise perceptually indistinguishable from hypoxia. We conclude that hypoxia and hypercapnia have equal potency for air hunger when matched by ventilatory drive. Air hunger may, therefore, arise via brain stem respiratory drive.