Electrocardiogram and intra-arterial blood pressure were recorded in % men (aged 35 to 45 years) by the Oxford method over a 30-hour period. The study involved 33 normotensive, 29 borderline hypertensive, and 34 mildly hypertensive individuals, as assessed by the cuff method. Five-minute periods during sleep and with subjects in supine, sitting, and standing positions were extracted from the recordings for frequency domain analysis. Power spectrum density estimates of systolic blood pressure, diastolic blood pressure, and heart rate were calculated by an autoregressive method over the bandwidths of 0.02 to 0.075 (low-frequency), 0.075 to 0.15 (midfrequency), and 0.15 to 035 Hz (high-frequency), attributable to thermoregulatory, baroreceptor, and respiratory activity. No significant intergroup differences were observed at nighttime, but in different body positions the borderline hypertensive subjects frequently had either greater low-frequency variability or smaller midfrequency variability than the other groups. In this respect, the power spectra for systolic and diastolic blood pressures provided better statistical differentiation between the groups than those for heart rate. Furthermore, the borderline hypertensive subjects exhibited attenuated night-day changes in the low-frequency band for all time series. The results suggest that in borderline hypertension the baroreceptor oscillations are shifted to lower frequencies, presumably reflecting altered function of the sympathetic nervous system. In conclusion, spectral analysis of blood pressure variability for controlled test situations made it possible to detect differences in the cardiovascular regulatory systems between normotensive, borderline hypertensive, and mildly hypertensive individuals. (Hypertension. 1994^3:18-24.) Key Words • blood pressure • hypertension, borderline • heart rate • spectrum analysis B y means of spectral analysis, short-term (time scale of seconds to minutes) variability in blood pressure (BP) and heart rate (HR) tracings can be divided into oscillatory components arising from various sources: one is linked to respiration (at approximately 0.25 Hz), 1 another is assumed to be caused by baroreceptor activity (at approximately 0.1 Hz), 2 and the third is believed to originate in the system responsible for regulating body temperature (in the range of 0.08 to 0.04 Hz).3 The slowest component also has been held to be due to local adjustments of resistance in individual vascular beds matching blood flow to local metabolic demand. 4 It is not known exactly how these fluctuations enter into the cardiovascular control loops, but they are generally considered to be mediated principally by the autonomic nervous system. Respiratory coupled variability, especially in HR, is a marker of vagal activity, whereas the lower frequencies are jointly mediated by parasympathetic and sympathetic outflows.56 Accordingly, spectral analysis is a method that can provide important information about autonomic nervous control