We investigated whether the greater degree of exercise-induced diaphragmatic fatigue previously reported in highly trained athletes in hypoxia (compared with normoxia) could have a contribution from limited respiratory muscle blood flow. Seven trained cyclists completed three constant load 5 min exercise tests at inspired O 2 fractions (F IO 2 ) of 0.13, 0.21 and 1.00 in balanced order. Work rates were selected to produce the same tidal volume, breathing frequency and respiratory muscle load at each F IO 2 (63 ± 1, 78 ± 1 and 87 ± 1% of normoxic maximal work rate, respectively). Intercostals and quadriceps muscle blood flow (IMBF and QMBF, respectively) were measured by near-infrared spectroscopy over the left 7th intercostal space and the left vastus lateralis muscle, respectively, using indocyanine green dye. The mean pressure time product of the diaphragm and the work of breathing did not differ across the three exercise tests. After hypoxic exercise, twitch transdiaphragmatic pressure fell by 33.3 ± 4.8%, significantly (P < 0.05) more than after both normoxic (25.6 ± 3.5% reduction) and hyperoxic (26.6 ± 3.3% reduction) exercise, confirming greater fatigue in hypoxia. Despite lower leg power output in hypoxia, neither cardiac output nor QMBF (27.6 ± 1.2 l min −1 and 100.4 ± 8.7 ml (100 ml) −1 min −1 , respectively) were significantly different compared with normoxia (28.4 ± 1.9 l min −1 and 94.4 ± 5.2 ml (100 ml) −1 min −1 , respectively) and hyperoxia (27.8 ± 1.6 l min −1 and 95.1 ± 7.8 ml (100 ml) −1 min −1 , respectively). Neither IMBF was different across hypoxia, normoxia and hyperoxia (53.6 ± 8.5, 49.9 ± 5.9 and 52.9 ± 5.9 ml (100 ml) −1 min −1 , respectively). We conclude that when respiratory muscle energy requirement is not different between normoxia and hypoxia, diaphragmatic fatigue is greater in hypoxia as intercostal muscle blood flow is not increased (compared with normoxia) to compensate for the reduction in P aO 2 , thus further compromising O 2 supply to the respiratory muscles.