This study aimed at comparing estimations of spontaneous cardiac baroreflex sensitivity (BRS) obtained with 3 different methods from continuous non-invasive blood pressure recordings in humans. A new method, allowing the quantification of the statistical dependence between values of 2 parameters (Z coefficient), was applied to beat-to-beat systolic blood pressure (SBP) and heart period (HP) values. SBP and HP values with positive Z coefficient and corresponding to baroreflex activity (SBP and HP values both lower or higher than the modal values) were submitted to a linear regression and the regression coefficient (Zgain) was taken as an index of BRS. Second, cross-spectral analysis of SBP and HP gave a BRS value (Csgain) computed as the average value of transfer function moduli for frequencies between 0.07 and 0.14 Hz, with coherence between SBP and HP greater than 0.5. The third method relies on the analysis of linear sequences (r > 0.97) containing at least 3 values of SBP and HP varying in the same direction. The average regression coefficient obtained from all selected SBP and HP sequences is the index of BRS (Seqgain). SBP and HR were recorded during 1 hour with a Finapres in 10 healthy male volunteers (NT), 23 to 32 year-old (SBP: 123 +/- 2 mmHg) and 10 recent and untreated hypertensive subjects (HT) (SBP: 152 +/- 6 mmHg). [table: see text] These results show that, in both groups, Zgain and Seqgain correlated with Csgain. No correlation was found between Zgain and Seqgain in healthy volunteers whereas the correlation was strong in hypertensives probably due to more heterogeneous SBP levels and BRS values in these subjects. This suggests that these methods are sensitive to different ways of response of the baroreflex.