Increasing hypoxia severity has been shown to exacerbate peripheral fatigue, which affects exercise performance during intermittent, maximal and submaximal isolated contractions. To date, how hypoxia severity modifies the neural versus muscular determinants of high-intensity, whole-body intermittent exercise to exhaustion is however unknown. The aim of this study was to examine the effects of hypoxia severity on quadriceps fatigability during exhaustive intermittent cycling. Fifteen well-trained cyclists performed an intermittent cycling exercise until exhaustion at supra-maximal intensity – 15 s at 30% of the anaerobic power reserve (609±23 W) with a fixed pedaling frequency of 110 rpm, interspersed with 45 s of passive rest – in normoxia (simulated altitude/end-exercise arterial O2 saturation = 0 m/96%), moderate (2200 m/90%) and severe hypoxia (4200 m/79%). Neuromuscular tests including electrical femoral nerve and transcranial magnetic stimulations during brief (5-s) and sustained (30-s) maximal isometric voluntary contractions (MVC) of the knee extensors were performed at baseline and 7 min post-exhaustion. Exercise performance differed (P<0.001) among the three conditions of oxygenation (39 ± 8, 22 ± 3 and 13 ± 2 sprint repetitions in normoxia, moderate and severe hypoxia, respectively). At exhaustion, the reduction in peak twitch amplitude (P<0.001) in response to supra-maximal, unpotentiated single (-53.2%) and potentiated paired (-33.1%) electrical stimuli was identical across conditions. Compared with baseline, strength loss during brief MVC was similar at exhaustion in normoxia and moderate hypoxia (-9.3% and -9.9%; both P<0.01), while a smaller (-6.3%; P=0.136) force decline occurred with severe hypoxia. When contraction was prolonged, exercise-induced decreases in force were consistent across conditions (-22.0% from the beginning to the end of the 30-s sustained MVC; all conditions compounded, P<0.01). This was accompanied by lower (P<0.05) end-exercise voluntary activation values obtained from both motor nerve and motor cortex stimulations during brief (-1.9% and -4.6%, respectively; all conditions compounded) and sustained (-2.1% and -9.7%, respectively; all conditions compounded) MVCs, with no hypoxia severity effect. Maximal M-wave amplitude (at rest and during MVCs) and maximal RMS activity of vastus lateralis and rectus femoris muscles obtained during MVCs did not change from pre- to post-exercise in all conditions. Despite earlier exercise cessation with increasing hypoxia severity, end-exercise reductions in quadriceps twitch force were similar across conditions. These results indicate the existence of a critical threshold of peripheral fatigue with intermittent cycling to exhaustion, independently of the hypoxia level. Another novel finding was that a suboptimal output from the motor cortex may also contribute to exhaustion.