We assessed the ability of the oxygen uptake efficiency slope, whether calculated on 100 and 80% of maximal exercise test duration (OUES(100) and OUES(80)), to identify the change in cardiorespiratory capacities in response to hypoxia in subjects with a broad range of V(O2 peak). Four maximal exercise tests were performed in trained (T) and untrained subjects (UT) in normoxia and at 1000, 2500 and 4500 m. The mean reductions in maximal exercise capacities at 4500 m were the same in T subjects for V(O2 peak) (-30%), OUES(80) (-26%) and OUES(100) (-26%) whereas in UT subjects only OUES(100) (-14%), but not OUES(80) (-20%), was lower compared with V(O2 peak) (-21%, p<0.05). OUES(100) and OUES(80) were correlated with V(O2 peak) and the ventilatory anaerobic threshold in both groups. Multiple regression analyses showed that V(O2 peak), OUES(100) and OUES(80) were significantly linked to O(2) arterial-venous difference. The OUES(80) could be considered as an interesting sub-maximal index of cardiorespiratory fitness in normal or hypoxemic subjects unable to reach V(O2 peak).