We thank the authors of the letter for their careful evaluation of our study (Grassi et al. 2005) and for the interesting points they have raised. The following are our responses to their comments. According to the letter our model (isolated dog gastrocnemius in situ) ‘might not be the ideal preparation for investigating VO2 kinetics’. Essentially all experimental models have disadvantages, and we are not aware of any ideal preparation for the study of VO2 kinetics. Despite some intrinsic limitations (discussed in our papers), our model offers several important advantages compared to conventional pulmonary VO2 kinetics analysis. One major advantage is the ability to directly determine VO2 across the exercising muscle, without the confounding effects of: (a) transit delays from the site where gas exchange occurs, that is skeletal muscles, and the site where it is determined, that is the mouth of horses/humans; and (b) changes of O2 stores between the two sites. Both (a) and (b) represent intrinsic and significant limitations in pulmonary VO2 kinetics analysis. Other advantages of our model are more specifically related to the effects of nitric oxide (NO) on VO2 kinetics. As discussed in our study, increases in the duration of the ‘cardiodynamic phase’ (which were seen in such studies as Kindig et al. (2001), Jones et al. (2003) and Wilkerson et al. (2004)), as well as vasoconstriction in venular beds, possible in the presence of NO synthase (NOS) inhibition, could produce the appearance of a faster pulmonary VO2 kinetics even in the presence of an unchanged muscle VO2 kinetics. Another advantage of our model, relevant for the study in question, is the capability of keeping muscle blood flow (Q) constantly elevated during the transition from rest to contractions. NO elicits vasodilatation and inhibits mitochondrial respiration. Thus, NOS inhibition could have opposing effects on VO2 kinetics: inhibition of vasodilatation could restrict O2 delivery and therefore slow VO2 kinetics, whereas relief of mitochondrial inhibition could speed the kinetics. By pump-perfusion, we kept Q constantly elevated, so that we were better suited (compared to conventional pulmonary VO2 kinetics analysis) to see effects of NOS inhibition on mitochondrial respiration. Nonetheless, we saw no effect on kinetics. The point about high perfusion pressure in our l-NAME condition, with the possibility of oedema and impairment of peripheral O2 diffusion, is well taken. However: (a) in the l-NAME condition perfusion pressure was elevated only for a brief period at rest, immediately prior to the onset of contractions; (b) perfusion pressures during the contraction periods were very similar for control and l-NAME conditions (see Table 2); (c) O2 extraction during contractions was similar in the two conditions (Table 2); and (d) there was no oedema, as demonstrated by an average percentage of water in the muscles (determined at the end of the experiment) of 75.8 ± 0.4, precisely in the range of values for one of our previous studies conducted without pump-perfusion (Grassi et al. 2002). We agree with the authors of the letter that any intramuscular QO2/VO2 maldistribution would work against O2 diffusion capacity and might affect O2 extraction and VO2 kinetics. In terms of QO2/VO2 maldistribution, however, it seems less likely in our model compared to more ‘normal’ experimental conditions. In our model, indeed: (a) all fibres are synchronously activated by electrical stimulation; (b) QO2 is kept constantly elevated; and (c) vasodilatation is assured by adenosine administration. All of these factors should improve QO2/VO2 maldistribution, providing better matching of blood flow and muscle metabolism. With regard to timing of arterial and venous sampling, arterial blood O2 concentration (CaO2) is essentially constant throughout our experimental periods. This makes timing of venous sampling relative to arterial sampling irrelevant. Nevertheless, it is an excellent point that application of the Fick equation is uncertain in any non-steady state. Strictly speaking, the Fick equation only applies when CaO2, venous O2 concentration, QO2 and metabolism are all constant (Zierler, 1961). In this respect our measures appear more appropriate than others, because we do have constant QO2 and CaO2. It should also be noted that similar concerns apply to the use of pulmonary gas exchange equations in the non-steady state. The point about fibre type distribution is well taken, and it was mentioned in the Discussion of our study. The dog gastrocnemius is a highly oxidative muscle, and we concur that the effects of NOS inhibition on VO2 kinetics might be more pronounced in muscles with a higher percentage of type 2 fibres, or during transitions to higher exercise intensities. The issues of the small muscle mass and of the single transition from rest to contractions, which (see Lamarra et al. 1987) would reduce confidence in parameter estimation for VO2 kinetics analysis, apply to breath-by-breath pulmonary VO2 measurements, and not to our model. It does not matter how small the muscle is, as long as VO2 can be reliably measured directly across it (as in our model), and not at a distant site (the lungs) with the confounding effects of the VO2 of the rest of the body. As shown in Fig. 2, the ‘noise’ of our VO2 measurements is much less than that observed with breath-by-breath pulmonary, even after superposition and averaging of data from multiple repetitions. We do not agree with the authors of the letter that ‘several of the individual model fits displayed in Fig. 2 of Grassi et al. (2005) could be considered questionable’: indeed, in 11 out of 12 of the individual figures the experimental data appear almost perfectly fitted by the single- or double-exponential equations. Again, this quality of fitting is higher than that usually obtained with breath-by-breath pulmonary VO2 measurements. For biopsy data, we agree with the authors of the letter that ‘superficial muscle samples might not necessarily reflect the overall muscle energetic state’. This applies to all biopsy measurements, which, however, are widely utilized in exercise physiology and can add information of critical importance compared to non-invasive studies. As for the comment ‘In a preparation that is not inherently stable’, we see no evidence to support this description of our preparation. The reduced muscle fatigue after l-NAME administration could be due to several effects of the drug (see the references cited in the study), independently from VO2 kinetics or O2 deficit. As for the trend toward lower estimates of substrate-level phosphorylation (SLP) in l-NAME, it could be explained in terms of a lower energy cost for force production (as discussed in our paper, and also suggested by the recent paper of Baker et al. 2006). In other words, the reduced muscle fatigue and lower SLP observed in our study after NOS inhibition do not necessarily mean a faster VO2 kinetics and a reduced O2 deficit, as postulated by the authors of the letter. When we defined the differences in VO2 kinetics, reported by the previous studies on horses/humans, as ‘rather small’ or ‘relatively small’ we were mainly referring to the almost indistinguishable average VO2 kinetics curves in the upper panel of Fig. 1 by Kindig et al. (2001), to Table 1 in the study by Jones et al. (2004) (small or very small differences in τ in 3–4 subjects out of 7), to Fig. 4 in the study by Wilkerson et al. (2004) (small or very small differences in τ in 3–4 subjects out of 7), to Table 2 in the study by Jones et al. (2003) (no or very small differences in τ in 3 subjects out of 7). However, we of course recognize that the observed differences were statistically significant. In Conclusion, we concur with the authors of the letter that the final sentence of our Abstract (Grassi et al. 2005) was perhaps too ‘bold’. However, we believe that our Conclusions paragraph was more cautious and therefore more correct. As discussed above and in the study, our experimental model has some intrinsic limitations but presents several important advantages compared to pulmonary VO2 kinetics analysis. Also, there are several possible factors that might explain our conflicting results as compared to data from horses (Kindig et al. 2001, 2002) or humans (Jones et al. 2003, 2004; Wilkerson et al. 2004). Our results provide evidence in favour of the concept that inhibition of mitochondrial respiration by NO may not limit VO2 kinetics, and recommend further studies in the area. We again commend the authors of the letter for pressing the discussion of NOS inhibition and VO2 kinetics. Such constructive debate will assist all investigators in this area as we search for the mechanisms underlying the adjustment of metabolism at the onset of rapid changes in energy demand.