New Findings: What is the central question of this study? Hyperoxia enhances endurance performance by increasing O 2 availability to locomotor muscles. We investigated whether hyperoxia can also improve prolonged self-paced exercise in conditions of elevated thermal and cardiovascular strain. What is the main finding and its importance? Hyperoxia improved self-paced exercise performance in hot and cool conditions. However, the extent of the improvement (increased work rate relative to normoxia) was greater in cool conditions. This suggests that the development of thermal and cardiovascular strain during prolonged self-paced exercise under heat stress might attenuate the hyperoxia-mediated increase in O 2 delivery to locomotor muscles. Abstract: The aim of this study was to determine whether breathing hyperoxic gas when self-paced exercise performance is impaired under heat stress enhances power output. Nine well-trained male cyclists performed four 40 min cycling time trials: two at 18°C (COOL) and two at 35°C (HOT). For the first 30 min, participants breathed ambient air, and for the remaining 10 min normoxic (fraction of inspired O 2 0.21; NOR) or hyperoxic (fraction of inspired O 2 0.45; HYPER) air. During the first 30 min of the time trials, power output was lower in the HOT (∼250 W) compared with COOL (∼273 W) conditions (P < 0.05). In the final 10 min, power output was higher in HOT-HYPER (264 ± 25 W) than in HOT-NOR (244 ± 31 W; P = 0.008) and in COOL-HYPER (315 ± 28 W) than in COOL-NOR (284 ± 25 W; P < 0.001). The increase in absolute power output in COOL-HYPER was greater than in HOT-HYPER (∼12 W; P = 0.057), as was normalized power output (∼30%; P < 0.001). The peripheral capillary percentage oxygen saturation increased in HOT-HYPER and COOL-HYPER (P < 0.05), with COOL-HYPER being higher than HOT-HYPER (P < 0.01). Heart rate was higher during the HOT compared with COOL trials (P < 0.01), as were mean skin temperature (P < 0.001) and peak rectal temperature (HOT, ∼39.5°C and COOL, ∼38.9°C; P < 0.01). Thermal discomfort was also higher in the HOT compared with COOL (P < 0.01), whereas ratings of perceived exertion were similar (P > 0.05). Hyperoxia enhanced performance during the final 25% of a 40 min time trial in both HOT and COOL conditions compared with normoxia. However, the attenuated increase in absolute and normalized power output noted in the HOT condition suggests that heat stress might mitigate the influence of hyperoxia.