RÉSUMÉ
In order to study respiratory transients during exercise, we examined breath-by-breath differences between gas exchange kinetics measured at the mouth and those estimated at the alveolar level. The gas exchange data at the mouth were obtained by measurement of expired gases only (expiratory flow method) . Correction for breath-by-breath changes in lung gas stores was applied to the total gas exchange, which was obtained by subtracting expired from inspired gas volume (alveolar gas exchange method) . Constant work loads (150, 200, 250 W) and a ramp work load (30 W/min) preceded and followed by a 50 W load were generated by a computerized cycle ergometer. Best-fit first- or second-order model values for gas exchange kinetic parameters were found by the non-linear least-squares method.<BR>1. Regardless of work intensity and forcing function, the breath-by-breath variation in gas exchange measured at the mouth was larger than the gas exchange estimated at the alveolar level, in both a non-steady state and a steady state. The variation was caused by the invalidity of assuming zero N<SUB>2</SUB> exchange at the mouth, which was attributed to changes in lung volume.<BR>2. Vo<SUB>2</SUB> kinetics at the alveolar level were faster than those at the mouth, while the converse held for Vco<SUB>2</SUB> at the onset of constant load work, due to the effects of fluctuations in lung gas stores on the kinetics of gas exchange at the mouth. During ramp load work, Vo<SUB>2</SUB> and Vco<SUB>2</SUB> kinetics at the alveolar level were faster than those at the mouth.<BR>3. Steady state gas exchange values at the alveolar level and at the mouth were the same during constant load work, since the lung gas stores corrections added up to small fractions of the total gas exchange when summed over the long term.<BR>4. Consideration of both the proper end-expiratory lung volume and ventilationperfusion inhomogeneity was required in order to estimate the true alveolar gas exchange.