Abstract : Persistence of respiratory sinus arrhythmia (RSA) has been described in humans during intense exercise and attributed to an increase in ventilation. However, the direct influence of ventilation on RSA has never been assessed. The dynamic evolution of RSA and its links to ventilation were investigated during exercise in 14 healthy men using an original modeling approach. An evolutive model was estimated from the detrended and high-pass-filtered heart period series. The instantaneous RSA frequency (FRSA, in Hz) and amplitude (ARSA, in ms) were then extracted from all recordings. A(RSA) was calculated with short-time Fourier transform. First, measurements of FRSA and ARSA were performed from data obtained during a graded and maximal exercise test. Influences of different ventilation regimens [changes in tidal volume (VT) and respiratory frequency (FR)] on ARSA were then tested during submaximal [70% peak O2 consumption (VO2peak)] rectangular exercise bouts. Under graded and maximal exercise conditions, ARSA decreased from the beginning of exercise to 61.9 +/- 3.8% VO2peak and then increased up to peak exercise. During the paced breathing protocol, normoventilation (69.4 +/- 8.8 l/min), hyperventilation (81.8 +/- 8.3 l/min), and hypoventilation (56.4 +/- 6.2 l/min) led to significantly (P < 0.01) different ARSA values (3.8 +/- 0.5, 4.6 +/- 0.8, and 2.9 +/- 0.5 ms, respectively). In addition, no statistical difference was found in ARSA when ventilation was kept constant, whatever the FR-VT combinations. Those results indicate that RSA persists for all exercise intensities and increases during the highest intensities. Its persistence and increase are strongly linked to both the frequency and degree of lung inflation, suggesting a mechanical influence of breathing on RSA.