In this study, the effects of long-term physical training on autonomic function in athletes and the response of the autonomic nervous system to dynamic exercise were investigated in nonathletes and athletes with power spectral analysis of heart rate variability (HRV). This study was performed on 13 healthy subjects (5 athletes and 8 non athletes). Electrocardiographic (ECG) signals were continuously recorded during (1) 15 min of rest in a sitting position on a bicycle ergometer, (2) the dynamic exercise test to the point of exhaustion, and (3) a 15 min postexercise period. After the recorded ECG signals were sampled at 500 samples/s, the instantaneous HRV signal was constructed from the detected R peaks and then resampled at 4 Hz in order to obtain an evenly spaced time series applicable to power spectral analysis. After linear trends were removed by the robust locally weighted regression algorithm, the power spectrum of HRV was estimated for contiguous records of 512 samples by Burg's maximum entrophy method. HRV was quantified by determining the spectral area (power) in two frequency bands, low-frequency power (LF power: 0.05-0.15 Hz) and high-frequency power (HF power: 0.15-0.8 Hz), and their ratio. The comparison between athletes and nonathlete was performed in terms of the above-mentioned parameters. Although both groups showed similar trends in heart rate (HR) at all stages of protocols, HR in athletes was significantly lower than that in nonathletes during rest and postexercise. In athletes and nonathletes, LF and HF powers gradually decreased with exercise. As recovery progressed, they continued to increase gradually, but remained below resting level. During rest and postexercise, HF power in athletes was significantly (p < 0.05) higher than than in nonathletes. Also, the recovery of HR and HF powers during early recovery (PO1) was more rapid in athletes than in nonathletes. Both groups showed an attenuation of LF and HF powers during dynamic exercise. It is likely that, in athletes, the lower HR during rest and the more rapid recovery of HR postexercise was due to a high level of HF power, indicating that vagal activity was enhanced by the adaptive changes in neural regulation produced by long-term physical training.