The study published by McLean et al. (2006) dealt with a very relevant topic. Actually, behind the question of the mechanisms involved in the improvement in performance following a living high-training low (LHTL) session, emerge the query of the suYcient dose of hypoxia insofar as one hypothesizes that oxygen carrying transport capacities are implicated in this improvement. However, we have some concerns mainly with the conclusions of the study of McLean and coworkers.Methodological concerns The authors compare their results to those obtained from the LHTL model; however subjects did not train during the study and were maybe even sedentary. It is far from unreasonable to think that training could have modiWed erythropoiesis. Moreover, an acute exposure of eight consecutive hours to an altitude of 4,925 m without any preliminary acclimatization have certainly triggered symptoms of acute mountain sickness (Richalet and Rathat 1991), but this is not reported in the article. Recovery after training would certainly be impeded by using such a high altitude. It has been recently proposed by Levine and Stray-Gundersen (in press) that an altitude below 3,000 m gives the best results in LHTL sessions. Lastly, for such a short exposure, a nocturnal exposure would be more eYcient that a diurnal one since it is known that the serum EPO level is at its zenith during night (Cahan et al. 1992).Markers of erythropoiesis To our mind the markers of erythropoiesis used by the authors are not the most accurate. Actually, total hemoglobin mass (Hbmass) measured by the CO-rebreathing method seems to be more pertinent since the variations in Hbmass are independent of circulating blood volume. Otherwise, the time course of Hct, Hb and RBC presented on Figs. 4, 5, and 6, respectively are very similar, which is not surprising. However what is remarkable is the very similar time course observed between control and hypoxic groups for these three parameters. The existing basal diVerence between groups persists during the whole study suggesting a limited eVect of the intervention. The signiWcant diVerence more suggests measurement artefacts rather than a physiological variation despite the absence of signiWcance among the control group on day 7.Is the proposition of the 34 UL -1 threshold in EPO secretion pertinent? In the conclusion, the authors speculate for a threshold EPO before any evidence of accelerated erythroid production. For the authors, this threshold is related to a daily exposure longer than 10 h per day. All the studies that have evidenced a role of the modiWcations of O 2 carrying capacities as a responsible factor for the improvement in aerobic performance did not report the level of EPO. In a work of our team ) the maximal level of EPO never exceeded half of the level proposed by McLean et al., but we have shown a clear stimulation of erythropoiesis as mentioned by the authors in this article. Paradoxically we even evidenced a higher Hbmass after 13 days of LHTL despite any increased EPO at the moment of the measurement ). Ne...