The relationship of respiratory sinus arrhythmia amplitude (RSA) to tidal volume and breathing frequency was quantified during voluntarily controlled tidal volume and breathing frequency and spontaneous quiet breathing. Seventeen seated subjects breathed via mouthpiece and nose-clip, maintaining constant tidal volumes at each of several breathing frequencies. Inspiratory breath hold was zero frequency. Log RSA was plotted vs. log frequency for each tidal volume. The large stable RSA for frequencies less than 6 cycles/min was called low-frequency intercept (LFI, 20 +/- 5 beats/min). Low-frequency intercept was inversely proportional to a subject's age only to 35 yr. At higher breathing frequencies above a characteristic corner frequency (fC, 7.2 +/- 1.5 cycles/min) RSA decreased with constant slope (roll-off; 21 +/- 3.4 dB/decade). The RSA-volume relationship was linear permitting normalization of RSA-frequency curves for tidal volume to yield one curve. Spontaneous breathing data points fell on this curve. Voluntarily coupling of heart rate to breathing frequency in integer ratios reduced breath-by-breath variability of RSA without changing mean RSA. In conclusion, low-frequency intercept, corner frequency, and roll-off characterize an individual's RSA-frequency relationship during both voluntarily controlled and spontaneous breathing.