• The hypothesis that the ventilatory resistance to O2 flow (RV) does limit maximal O2 consumption (V˙O2 max) in hypoxia, but not in normoxia, at least in non-athletic subjects, was tested. RV was reduced by using He–O2 mixtures. • V˙O2, max was measured during graded cyclo-ergometric exercise in eight men (aged 30 ± 3 years) who breathed N2–O2 and He–O2 mixtures in normoxia (inspired oxygen fraction (FI,O2= 0.21) and hypoxia (FI,O2= 0.11). O2 consumption, expired and alveolar ventilations (V˙E and V˙A, respectively), blood lactate and haemoglobin concentrations, heart rate and arterial oxygen saturation (Sa,O2) were determined at the steady state of each work load. Arterial O2 and CO2 partial pressures (Pa,O2) and Pa,CO2, respectively) were measured at rest and at the end of the highest work load. • Maximal V˙E and V˙A were significantly increased by He—O2 breathing in normoxia (+27 and +18%, respectively), without significant changes in Pa, O2, Sa, O2V˙O2, max In hypoxia, V˙E and V˙A increased (+31 and +24%, respectively), together with Pa,O2 (+17%), Sa, O2 (+6%) and V˙O2, max (+14%). • The results support the hypothesis that the role of RV in limiting V˙O2,max is negligible in normoxia. In hypoxia, the finding that higher V˙E and V˙A values during He–O2 breathing led to higher V˙O2,max values suggests a greater role of RV as a limiting factor. It is unclear whether the finding that the V˙O2 max values were the same during He–O2 and N2–O2 breathing in normoxia is due to a non-linear response of the O2 transfer system, as previously proposed.