Résumé : Ballistocardiography (BCG), measuring the heart beat-induced mass movements of the body generated by forces associated to heart contraction, has been recently reconsidered as a useful non-invasive technique to characterize cardiac performance, potentially offering a simple, efficient and affordable solution for e-health monitoring. Our aim was to extract heart rate variability indices from BCG signal (B-HRV), and to evaluate their correlation with ECG-derived (E-HRV) ones, both in time and frequency domains. Five-minute BCG signals from 18 healthy volunteers were acquired by a piezoelectric 3D-force plate simultaneously with the ECG signal, in both standing (ST) and supine (SUP) positions. BCG systolic waves (H, I, J and K) in the longitudinal (head-foot) component were detected, from which B-HRV series were computed and compared by correlation and Bland-Altman analysis against conventional RR. Posture affected accuracy of B-HRV time and frequency parameters compared to ECG-HRV, with the best performance obtained in SUP by the JJ series. This signal could be used as an alternative measurement of HRV in novel home e-health monitoring systems.