Several studies have demonstrated that hyperoxia increases the maximal O
2
consumption rate (ṀO
2max
) in fish, but exactly how this occurs remains to be explained. Here, we tested the hypothesis that hyperoxia improves arterial oxygenation in rainbow trout during exhaustive exercise. We demonstrate a 35% higher ṀO
2max
in hyperoxia (200% air saturation) relative to normoxia, which was achieved through a combined 15% increase in cardiac output due to elevated peak heart rate, and a 19% increase of the arterial–venous (A-V) O
2
content difference. While arterial O
2
partial pressure (PaO
2
) and O
2
saturation of haemoglobin declined post-exhaustive exercise in normoxia, this did not occur in hyperoxia. This protective effect of hyperoxia on arterial oxygenation led to a 22% higher arterial O
2
content post-exhaustive exercise, thereby allowing a higher A-V O
2
content difference. These findings indicate that ṀO
2max
is gill diffusion limited in exhaustively exercised rainbow trout. Moreover, as previous studies in salmonids have demonstrated collapsing PaO
2
in normoxia at maximal swimming speed and at acutely high temperatures, a diffusion limitation may constrain ṀO
2
in other situations eliciting peak metabolic demand. These findings, along with the fact that hyperoxia increases ṀO
2max
in several other fishes, suggest that gill diffusion limitations of ṀO
2max
may be widespread in fishes.