It remains unclear by which mechanism ‘live high–train low’ (LHTL) altitude training increases exercise performance. Haematological and skeletal muscle adaptations have both been proposed. To test the hypotheses that (i) LHTL improves maximal oxygen uptake (VO₂max) and (ii) this improvement is related to hypoxia-induced increases in total haemoglobin mass (Hb_mass) and not to improved maximal oxidative capacity of skeletal muscle, we determined VO₂max before LHTL and after LHTL, before and after the altitude-induced increases in Hb_mass (measured by carbon-monoxide rebreathing) had been abolished by isovolumic haemodilution. We obtained skeletal muscle biopsies to quantify mitochondrial oxidative capacity and efficiency. Sixteen endurance-trained athletes were assigned (double-blinded, placebo controlled) to ≥16 h/day over 4 weeks to normoxia (placebo, n=6) or normobaric hypoxia equivalent to 3000 m altitude (LHTL, n=10). Four-week LHTL did not increase VO₂max, irrespective of treatment (LHTL: 1.5%; placebo: 2.0%). Hb_mass was slightly increased (4.6%) in 5 (of 10) LHTL subjects but this was not accompanied by a concurrent increase in VO₂max. In the subjects demonstrating an increase in Hb_mass, isovolumic haemodilution elicited a 5.8% decrease in VO₂max. Cycling efficiency was altered neither with time nor by LHTL. Neither maximal capacity of oxidative phosphorylation nor mitochondrial efficiency was modified by time or LHTL. The present results suggest that LHTL has no positive effect on VO₂max in endurance-trained athletes because (i) muscle maximal oxidative capacity is not improved following LHTL and (ii) erythrocyte volume expansion after LHTL, if any, is too small to alter O₂ transport.