In this study, we hypothesized that 5 weeks of cycling endurance training can decrease the magnitude of the non-proportional increase in oxygen uptake (˙ V O 2) to power output relationship (˙ V O 2 'excess') at exercise intensities exceeding the lactate threshold (LT). Ten untrained, physically active men performed a bout of incremental cycling exercise until exhaustion before and after training. The mitochondrial DNA copy number, myosin heavy chain composition and content of uncoupling protein 3 and sarcoplasmic reticulum Ca 2+-ATPases (SERCAs) were analysed in muscle biopsies taken from vastus lateralis before and after training. The training resulted in an enhancement of the power-generating capabilities at maximal oxygen uptake (˙ V O 2 max) by ∼7% (P = 0.002) despite there being no changes in ˙ V O 2 max (P = 0.49). This effect was due to a considerable reduction in the magnitude of the ˙ V O 2 'excess' (P < 0.05) above the LT. A decrease in plasma ammonia concentration was found during exercise after training (P < 0.05). A downregulation of SERCA2 in vastus lateralis (P = 0.006) was observed after training. No changes in myosin heavy chain composition, selected electron transport chain proteins, uncoupling protein 3 or the mitochondrial DNA copy number (P > 0.05) were found after training. We conclude that the training-induced increase in power-generating capabilities at ˙ V O 2 max was due to attenuation of the ˙ V O 2 'excess' above the LT. This adaptive response seems to be related to the improvement of muscle metabolic stability, as judged by a lowering of plasma ammonia concentration. The enhancement of muscle metabolic stability after training could be caused by a decrease in ATP usage at a given power output owing to downregulation of SERCA2 pumps. The oxygen cost of cycling above the lactate threshold (LT) can be decreased after a few weeks of endurance training, as expressed by the lowering of the magnitude of the slow component of oxygen uptake (˙ V O 2) kinetics (Casaburi et al. 1987; Womack et al. 1995; Carter et al. 2000). However, the mechanism responsible for this effect remains unclear. There is a growing body of evidence to suggest that the increase in the oxygen cost of work during heavy-intensity exercise is caused by fatigue of the active muscle fibres (Zoladz et al. 2008; Hepple et al. 2010; Cannon et al. 2011) and, in particular, by a disturbance in the concentrations of muscle metabolites, i.e. a decrease in muscle phosphocreatine and a rise in free ADP, free AMP, free inosine monophosphate, creatine, inorganic phosphate and H + , potent to decrease muscle mechanical efficiency (Woledge, 1998; Zoladz et al. 2006). It is well known that endurance training can enhance muscle metabolic stability, i.e. lead to reduced changes in the concentrations of the above-mentioned muscle metabolites for a given ATP turnover (approximately