Although in physiological conditions RR interval and systolic arterial pressure (SAP) are likely to interact in a closed loop, the traditional cross-spectral analysis cannot distinguish feedback (FB) from feedforward (FF) influences. In this study, a causal approach was applied for calculating the coherence from SAP to RR (Ks-r) and from RR to SAP (Kr-s), and the gain and phase of the baroreflex transfer function. The method was applied, in comparison with the noncausal one, to RR and SAP series taken from 15 healthy young subjects in the supine position and after passive headup tilt. For the low frequency (0.04-0.15 Hz) spectral component, the enhanced FF coupling (Kr-s=0.59±0.21, significant in 14 subjects) and the blunted FB coupling (Ks-r=0.17±0.17, significant in 4 subjects) found at rest indicated the prevalence of nonbaroreflex mechanisms. The tilt manoeuvre recovered FB influences (Ks-r=0.47±0.16, significant in 14 subjects), which were stronger than FF interactions (Ks-r=0.34±0.19, significant in 9 subjects). At the respiratory frequency, the RR-SAP regulation was balanced at rest (Ks-r=0.30±0.18, Kr-s=0.29±0.20, significant in 11 and 8 subjects) and shifted towards FB mechanisms after tilt (Ks-r=0.35±0.19, Kr-s=0.19±0.11, significant in 14 and 8 subjects). The causal baroreflex gain estimates were always lower than the corresponding noncausal values, and decreased significantly from rest to tilt in both frequency bands. The tilt-induced increase of the phase lag from SAP to RR suggested a shift from vagal to sympathetic modulation. Thus, the importance of nonbaroreflex interactions pointed out the necessity of accounting for causality in the cross-spectral analysis of the interactions between cardiovascular variables in healthy humans.