After local ethical approval, the kinetics of O₂ delivery ( Q_a,O2) at exercise onset was determined on four men (age 25.5 ± 4.2 years, _O2,max 4.42 ± 0.65 l min^-1), who performed four rest-to-50 W and rest-to-100 W transitions on a cycle ergometer. Beat-by-beat cardiac output ( Q) was measured by the model flow method (Wesseling et al. 1993) from continuously recorded pulse pressure profiles, calibrated against the open circuit acetylene method (Barker et al. 1999) at rest and at the exercise steady state. The trials were superimposed and averaged. Arterial O₂ saturation (S_a,O2) was measured by infrared oximetry, and the tracings superimposed and averaged. Blood haemoglobin concentration (Hb) was measured every minute and temporally aligned. Both parameters were interpolated by a 6th degree polynomial. Beat-by-beat Q_a,O2 was calculated as Q X S_a,O2 X Hb X 1.34 (O₂ binding coefficient). The half-times of the Q_a,O2 kinetics were 5.52 ± 2.46 and 7.33 ± 2.03 s in the rest-to-50 W and rest-to-100 W transients, respectively. The steady-state Q_a,O2 was proportional to power, and showed oscillations around its mean value. The time constant of the Q_a,O2 kinetics, as observed during light exercise, is lower than that reported for the _O2 kinetics in conditions of no lactate accumulation during the exercise transient (Binzoni et al. 1992; di Prampero & Ferretti, 1999), indicating that the Q_a,O2 kinetics is faster than the _O2 kinetics. This reflects the rapid Q response by the Frank-Starling mechanism (De Cort et al. 1991) and implies O₂ store changes during the exercise transient. The oscillations at steady state reflect the cardiorespiratory tuning of Q_a,O2.

This study was supported by the Swiss National Science Foundation, grant number 32-61780.00.

All procedures accord with current local guidelines and the Declaration of Helsinki.