To assess the relative roles of neural and nonneural mechanisms in respiratory sinus arrhythmia (RSA) at rest and during exercise (steady-state supine cycle ergometry at 25% of peak oxygen uptake), we studied 10 healthy men (mean age 21 +/- 1 yr) before (control) and during ganglion blockade (GB) with trimetaphan camsylate (3-5 mg/min i.v.). GB was confirmed by the abolition of the reflex bradycardia in response to intravenous phenylephrine and of the blood pressure rise with the cold pressor test. RSA was calculated from the power of the spectral component of the R-R interval variability centered at the breathing frequency. GB decreased but did not abolish RSA. At rest, this nonneural component of RSA was negligible, accounting for < 1% of the control RSA. During GB, exercise did not affect RSA significantly. However, because control RSA was decreased by exercise, the proportion of nonneural RSA increased by 32% (range from 17 to 75%). These results indicate that as the vagal tone decreases with exercise, an increasing proportion of RSA is due to nonneural mechanisms.