The present study proposed procedure for predicting an optimal left and right ventricular pacing interval delay (V-V interval). In 16 patients (heart failure, left bundle branch block, biventricular pacing) two methods (A and B) identifying optimal V-V interval were tested. Method A: predicted optimal V-V interval A (POVV-A) = electromechanical delay of the segment paced by left ventricle lead minus electromechanical delay of the segment paced by right ventricle lead. Method B: predicted optimal V-V interval B (POVV-B) = difference in the onset of aortic and pulmonary flows. Both methods were validated using echocardiography and right-sided heart catheterization. Cardiac output during POVV-A (4.6 l.min(-1)) was significantly better than that during POVV-A minus 20 ms (4.3 l.min(-1), p<0.01) and POVV-A plus 20 ms (4.3 l.min(-1), p<0.01), and than that during POVV-B (4.4 l.min(-1), p<0.05). LV dP/dt during POVV-A (818 mm Hg.s(-1), exceeded that during POVV-A plus 20 ms (717 mm Hg.s(-1),, p<0.05) and POVV-A minus 20 ms (681 mm Hg.s(-1), p<0.05), and that during POVV-B (727 mm Hg.s(-1), p<0.01). The time difference in onsets of myocardial deformation of left ventricle segment paced by the left ventricle and right ventricle lead allows identifying the optimal V-V interval and improves left ventricle performance.