Exercise in normal human subjects causes deterioration of matching of ventilation to blood flow in the lungs, but only in about 50% of those examined. A previous study (Wagner et al. 1989) of 5 horses showed no significant worsening of ventilation/blood flow (V̇a/Q̇) relationships during heavy exercise as determined by multiple inert gas elimination technique (MIGET). Because of the small number of horses in that study and the 50% human incidence of exercise induced V̇a/Q̇ mismatch, we studied an additional 6 Thoroughbreds, comparing V̇a/Q̇ relationships at the walk (1.4 m/s, 0o incline) and during galloping (9.6 ± 0.3 m/s, 7% incline). Such data were collected under 4 different conditions wherein inspired gas was 1) air, 2) 21% O2 in helium, 3) 15% O2 in N2 and 4) 15% O2 in helium. Each horse exercised 4 times (morning and afternoon of 2 days, with inspired gas conditions randomised). There was a small but significant increase in V̇a/Q̇ mismatch (similar under all 4 conditions). The second moment of the V̇a/Q̇ distribution (determined by the MIGET) increased significantly (P < 0.01) from 0.31 ± 0.01 at the walk to 0.38 ± 0.02 during gallop. This increase however is small: 0.38 is well within the range of this parameter for normal human subjects (where the 95% upper confidence limit is 0.60). This study shows that a small amount of exercise induced V̇a/Q̇ mismatch can occur in the horse as in man, but the mechanism remains to be elucidated and its clinical significance remains to be established. It may reflect interstitial pulmonary oedema but effects on arterial oxygenation are minimal.