To validate power spectral analysis of heart rate variability (HRV) as an autonomic indicator during exercise, ten males performed four identical progressive cycling tests during infusions of saline, esmolol (beta 1-blocker), glycopyrrolate (muscarinic blocker), or both drugs. Power spectra were constructed from the recorded electrocardiogram by Fourier algorithm and integrated for low-frequency power (LF) and high-frequency power (HF). Four different LF bands (0.004-0.1, 0.004-0.15, 0.05-0.1, and 0.05-0.15 Hz) and two different HF bands (0.1-1.0 and 0.15-1.0 Hz) were evaluated. The parasympathetic index, HF, decreased exponentially with workload and was attenuated by glycopyrrolate and combined treatments with both HF frequency bands measured. Whereas some sympathetic indexes (LF/total power and LF/HF) did reflect expected increases in sympathetic nerve activity associated with progressive increases in work intensity, none of the measured increases responded appropriately to autonomic blockade. It is concluded that HRV is a valid technique for noninvasive measurement of parasympathetic tone during exercise, but its validity as a measure of sympathetic tone during exercise is equivocal.