Disturbed sympathetic and also parasympathetic activity of the autonomic nervous system points to pathological alterations in the cardiovascular system. Untreated hypertensive subjects were examined with respect to the question of whether an increase in sympathetic activity necessarily goes along with a reduction in parasympathetic activity, and whether 'delayed' recovery behaviour after mental stress could be an indicator of a disturbed cardiovascular function. In 20 male hypertensive (HT) and 20 normotensive (NT) subjects (control group), heart rate variability (HRV) was compared during rest and under mental stress. The testing procedure consisted of the following phases: habituation, arithmetic tasks without and with interference, recovery. HRV was analysed using the trigonometric regressive spectral analysis (TRS). Proceeding from the total variance (ms2), the weighted averaged frequency (Hz) and the variance parts (ms2) in the frequency bands 'low frequency' (LF-band: 0.04-0.15 Hz) and 'high frequency' (HF-band: 0.15-0.40 Hz) were explored. The variance part modulated by spontaneous breathing within the HF-band was assessed additionally. The variance part in the LF-band under mental stress was significantly increased in the HT group (P<0.01). Activity in the HF-band (without the respiration-dependent part) under mental stress did not differ between both groups, whereas the breathing-modulated part of variance in the HF-band was reduced in the HT subjects. During the recovery period in the HT group, the weighted averaged frequency was still elevated compared to baseline, and the variance part in the LF-band was increased, which may point to delayed recovery behaviour. In addition, by using a discriminant analysis 85% of all subjects were reclassified to the original groups, all HT subjects being assigned 'correctly'. Spectral variance parameters enable early discovery of altered cardiovascular regulation. Respiration influences variance in the HF-band in hypertensive subjects and should therefore be paid attention to. The variance part in the LF-band, weighted averaged frequency and the respiration-modulated variance in the HF-band turned out to be the most valid parameters for the differentiation between NT and HT subjects.