Breathing increases abruptly at the start of passive exercise, stimulated by afferent feedback from the moving limbs, and declines towards a steady-state hyperpnoea as exercise continues. This decline has been attributed to both decreased arterial carbon dioxide levels, and adaptation in afferent feedback, however the relative importance of these two mechanisms is unknown. To address this issue, we compared ventilatory responses to 5 minutes of passive leg extension exercise performed on 10 awake human subjects (6 male, 4 female) in both isocapnic and poikilocapnic conditions. End-tidal carbon dioxide levels decreased significantly during poikilocapnic (=-1.5±0.5 mmHg, p<0.001), but not isocapnic passive exercise. Despite this difference, the ventilatory responses to passive exercise were not different between the two conditions. Using the fast changes in ventilation at the start (5.46±0.40 l^.min^-1, p < 0.001) and end (3.72±0.33 l^.min^-1, p < 0.001) of passive exercise as measures of the drive to breathe from afferent feedback we found that it declined by 68%. We conclude that the decline in ventilation during passive exercise is due to an adaptation in the afferent feedback from the moving limbs, not a decline in carbon dioxide levels.