It remains uncertain as how the reduction in systemic oxygen transport limits high-intensity exercise tolerance. 11 participants (5 males; age 35 ± 10 years; peak [Formula: see text] 3.5 ± 0.4 L min(-1)) performed cycle ergometry to the limit of tolerance: (1) a ramp test to determine ventilatory threshold (VT) and peak [Formula: see text]; (2) three to four constant-load tests in order to model the linear P-t (-1) relationship for estimation of intercept (critical power; CP) and slope (AWC). All tests were performed in a random order under moderate hypoxia (FiO(2) = 0.15) and normoxia. The linearity of the P-t (-1) relationship was retained under hypoxia, with a systematic reduction in CP (220 ± 25 W vs. 190 ± 28 W; P < 0.01) but no significant difference in AWC (11.7 ± 5.5 kJ vs. 12.1 ± 4.4 kJ; P > 0.05). However, large individual variations in the change of the latter were observed (-36 to +66%). A significant relationship was found between the % change in CP (r = 0.80, P < 0.01) and both peak [Formula: see text] (CP: r = -0.65, P < 0.05) and VT values recorded under normoxia (CP: r = -0.65, P < 0.05). The present study demonstrates the aerobic nature of the intercept of the P-t (-1) relationship, i.e. CP. However, the extreme within-individual changes in AWC do not support the original assumption that AWC reflects a finite energy store. Lower hypoxia-induced decrements in CP were observed in aerobically fitter participants. This study also demonstrates the greater ability these participants have to exercise at supra-CP but close to CP workloads under moderate hypoxia.