Multiple treadmill-based studies using low numbers of horses have evaluated potential prophylactic treatments for exercise-induced pulmonary haemorrhage (EIPH) and found no effect. However, the relevance of these findings to racing is unclear. Because severity of EIPH incurred on treadmills has not been compared to that following high-speed racetrack exercise in the same horses, we retrospectively performed this comparison using bronchoalveolar lavage fluid red cell numbers (BALFRBC) due to the relative insensitivity of tracheobronchoscopy. Six race-fit Thoroughbreds with recent tracheobronchoscopic EIPH scores ≥2 were exercised to fatigue on a treadmill at 115% V̇O2max (5% incline, 12.3-14.2 m/s), and maximally on a racetrack over 800 m and 1,100 m with average speeds ranging from 16.4-16.7 and 15.5-16.6 m/s, respectively. Run order varied but was not randomised. Bronchoalveolar lavage (BAL) was performed blindly using Bivona tubes 45-60 mins post-exercise. BALFRBC were determined using a haemocytometer. Data were expressed as median and interquartile range, and analysed using RM ANOVA with significance set at P<0.05. BALFRBC were greater after both racetrack runs than after treadmill exercise (P<0.05; treadmill: 10,305/μl (3,871-26,079); 800m: 25,000/μl (17,175-73,400); 1,100m: 19,500/μl (8,962-800,600). Treadmill exercise resulted in lower numbers and a narrower range in BALFRBC than racetrack exercise. Thus, when a small number of horses is used to study EIPH treatments on a treadmill, a lower BALFRBC would be anticipated following the baseline run than with a similar study using racetrack exercise, and might reduce the likelihood of demonstrating significant treatment effects. Results of this retrospective study raise concern regarding the advisability of extrapolating conclusions regarding efficacy of EIPH treatments from treadmill studies to racetrack scenarios.