The primary goal was to investigate the relationship between dynamic upper airway abnormalities and arterial blood gas tensions during exercise. Horses that completed a high-speed treadmill examination consisting of upper-airway videoendoscopy, blood gas evaluation and electrocardiogams and, postexercise, echocardiograms and tracheal washes, were included. An age-matched group of fit, healthy Thoroughbreds, trained to run on a high-speed treadmill, served as controls for blood gas values at specific exercise speeds. One hundred and nineteen horses completed the treadmill examination. Sixty (50%) were Thoroughbreds (TB), 51 (43%) Standardbreds (STD) and 8 (7%) other breeds. Mean +/- s.d. age TB 3.8 +/- 2.2 years and STD 4.0 +/- 1.7 years, with no gender predilection. Fifty-four horses (45%) had abnormal upper respiratory tract (URT) abnormalities alone or in combination with abnormalities in another body system. Thirty-eight (70%) were TB, 14 (26%) were STD and 2 (4%) were other breeds. Of these, 24 (45%) had exercising PaO2 values significantly lower than those observed in healthy TB. Nineteen (35%) horses also had significantly elevated exercising PaCO2. Only 14 (12%) horses had abnormal clinical findings in the URT alone, and of these, only 3 (21%) had an abnormally low PaO2 and/or elevated PaCO2. Multiple URT abnormalities were more commonly associated with abnormal exercising blood gases than were single disorders, but pharyngeal collapse (PC) was much more commonly associated with abnormal values if only one disorder was detected. Fifty-five percent (n = 65) of all cases admitted had no evidence of URT disease. Twenty-two (35%) were TB and 37 (57%) were STD. Twenty (31%) of these had abnormally low PaO2 and 14 (22%) had elevated PaCO2 values. Seventy percent (14) of the horses with abnormal PaO2 were STD, while almost 80% (11) of the horses with elevated PaCO2 were STD. These data suggest that dynamic URT dysfunction can adversely affect gas exchange during exercise. While multiple abnormalities were more commonly associated with gas exchange problems than were single disorders, pharyngeal collapse, either alone or in combination with other URT problems, was the disorder most frequently associated with blood gas abnormalities. Additionally, URT disease was more commonly seen in TB, and the proportion of URT diagnoses in horses with abnormal blood gases reflected this percentage, while STD without URT disease had a much higher incidence of abnormal blood gases than did TB without URT abnormalities.