Horses exercising at > or = approximately 90% VO2max develop arterial hypoxaemia with concurrent hypercapnoea, whereas ponies exercising at comparative levels become hypocapnoeic and maintain arterial oxygen tensions close to resting values. We sought to investigate the possibility that these differences relate to the ventilatory responses of these animals to exercise. Six Thoroughbred horses weighing mean +/- s.e. 501 +/- 27 kg and 5 ponies weighing mean +/- s.e. 164 +/- 18 kg exercised for 2 min on a 10% slope at speeds calculated to require 60% VO2max and for at least 1 min at speeds calculated to require 115% VO2max. Oxygen consumption (VO2), arterial oxygen (PaO2) and carbon dioxide (PaCO2) tensions, acid-base balance, tidal volume (VT), minute ventilation (VE), peak inspiratory (VImax) and expiratory (VEmax) flow, and maximal changes in transpulmonary pressures (delta PtPmax) were measured immediately before exercise and in the last 15 s of exercise at each intensity. The results confirmed that, unlike horses, ponies do not become hypoxaemic or hypercapnoeic during exercise. Despite having a higher delta PtPmax, higher VImax and VEmax and VE/kg0.75 at the same relative intensities, horses were less capable of mounting an appropriate ventilatory response to exercise. This was reflected by lower mass specific and metabolic weight-based ventilations at similar absolute workloads, and their higher PaCO2 and arterial HCO3-[, and lower ventilatory equivalent for oxygen (VE/VO2). This suggests that horses become hypoxaemic and hypercapnoeic at work loads > or = approximately 90% VO2max because their metabolic demand surpasses the capacity of their ventilatory system to meet this demand. Because ponies are less capable athletes, they can match their ventilatory response to their metabolic requirements.