Changes in dead space can explain part of the reduction in gas exchange efficiency found, not necessarily linked to respiratory sinus arrhythmia That respiratory sinus arrhythmia (RSA) improves pulmonary gas exchange and circulatory efficiency is an intriguing hypothesis (Yasuma & Hayano, 2004). J. Hayano and colleagues pioneered the research in this area and suggested that RSA-related changes in physiological dead space to tidal volume ratio (V D,phys /V T ) and the fraction of intrapulmonary shunt could be the links between RSA and efficiency of gas exchange, given that in a canine model both parameters increase when RSA is present under negative pressure ventilation produced by diaphragm pacing and under RSA changes induced by direct vagal stimulation (Hayano et al. 1996). From this point of view, the paper of Ito et al. (2006) is a necessary and valuable effort to investigate to what extent such a suggestion can be extended to humans during spontaneous ventilation and RSA. The results presented by the authors clearly indicate that vagal blockade using atropine results in attenuation of RSA, increase of V D,phys (mainly the anatomical dead space, V D,an ) and deterioration of pulmonary oxygenation. It would seem natural to take such evidence as confirmation that changes in V D,phys are likely to be one of the mechanisms linking variation of RSA with those of pulmonary oxygenation.However, in the Discussion, the authors state that 'the deterioration of pulmonary oxygenation after atropine administration in the present study appears to result from the mechanism of an increase in venous admixture associated with a decrease in RSA magnitude' and also report that 'the reduction in alveolar ventilation by the increase in physiological dead space [. . .] results in a negligible calculated reduction in P aO2 (<0.01 mmHg)' [is the arterial partial pressure of O 2 ]. To critically evaluate this last statement, we performed simulations using mathematical models available in the literature. The results found do not support the interpretation of the authors that the reduction found in V D,phys did not play a role in worsening pulmonary oxygenation. To explain our point, we briefly report the methods used and the outcomes of the simulations. The simulations were based on the unicompartmental model proposed by Ursino et al. (2001). We eliminated the part of the model performing ventilation control, since respiration was paced, and the part simulating blood flow regulation (imposing a constant value of 5 l min -1 ), to eliminate possible sources of P aO2 changes other than V D,phys . Given the specific objectives of the investigation, we added to the overall model a model of alveolar flow and volume during spontaneous ventilation (an adaptation of that presented in Section 2.3 of Khoo (1999) was used), an alveolar dead space in parallel with the alveolar compartment (modelled as a gas mixer, as the alveolar compartment, but unperfused), and an anatomical dead space (modelled as a series of gas mixers as, for example, in Ha...