Autonomic nervous system control in subjects with vasovagal syncope is controversial. In the present study, we used short-term spectral analysis to evaluate autonomic control in subjects with recurrent vasovagal syncope. We assessed the ability of spectral indices of HR (heart rate) variability to predict tilt-test responses. A series of 47 outpatients with recurrent vasovagal syncope and with positive responses to head-up tilt testing underwent a further study of RR variability during controlled breathing at rest and during tilt testing. During controlled breathing, RR interval variability of total power (TP(RR); P<0.001), low-frequency power (LF(RR); P<0.05), high-frequency power (HF(RR); P<0.001) and HF expressed in normalized units (HFnu(RR); P<0.001) were all higher, and LF expressed in normalized units (LFnu(RR)) and LF/HF ratio were lower in subjects with vasovagal syncope than in controls (P<0.001). To assess the ability of spectral components of RR variability to predict tilt-test responses, we prospectively studied 109 subjects with recurrent vasovagal syncope. The two normalized measures, HFnu(RR) and LFnu(RR), determined during controlled breathing alone predicted a positive tilt-test response (sensitivity, 76%; specificity, 99%; positive predictive value, 96%; and negative predictive value, 90%). During tilting, subjects with vasovagal syncope had lower SBP (systolic blood pressure; P<0.05), LF component of peak SBP variability (LF(SBP)) and LFnu(RR) than controls, and higher TP(RR), HF(RR), HFnu(RR) and alpha HF (P<0.001). These spectral data indicate that vagal sinus modulation is increased at rest in subjects with vasovagal syncope. Spectral analysis of RR variability during controlled breathing, a procedure that predicts tilt-test responses, could be a useful guide in choosing the method of tilt testing.