1. It is often assumed that the power in the low- (around 0.10 Hz) and high-frequency (around 0.25 Hz) bands obtained by power spectral analysis of cardiovascular variables reflects sympathetic and vagal tone [corrected] respectively. An alternative model attributes the low-frequency band to a resonance in the control system that is produced by the inefficiently slow time constant of the reflex response to beat-to-beat changes in blood pressure effected by the sympathetic (with or without the parasympathetic) arm(s) of the baroreflex (De Boer model). 2. We have applied the De Boer model of circulatory variability to patients with varying baroreflex sensitivity to patients with varying baroreflex sensitivity and one normal subject, and have shown that the main differences in spectral power (for both low and high frequency) between and within subjects are caused by changes in the arterial baroreflex gain, particularly for vagal control of heart rate (R-R interval) and left ventricular stroke output. We have computed the power spectrum at rest and during neck suction (to stimulate carotid baroreceptors). We stimulated the baroreceptors at two frequencies (0.1 and 0.2 Hz), which were both distinct from the controlled respiration rate (0.25 Hz), in both normal subjects and heart failure patients with either sensitive or poor baroreflex control. 3. The data broadly confirm the De Boer model. The low-frequency (0.1 Hz) peak in either R-R or blood pressure variability) was spontaneously generated only if the baroreflex control of the autonomic outflow was relatively intact.(ABSTRACT TRUNCATED AT 250 WORDS)