The goal of the present study was to assess the effects of left ventricular (LV) pacing sites (apex vs. free wall) on radial synchrony and global LV performance in a canine model of contraction dyssynchrony. Ultrasound tissue Doppler imaging and hemodynamic (LV pressure-volume) data were collected in seven anesthetized, opened-chest dogs. Right atrial (RA) pacing served as the control, and contraction dyssynchrony was created by simultaneous RA and right ventricular (RV) pacing to induce a left bundle-branch block-like contraction pattern. Cardiac resynchronization therapy (CRT) was implemented by adding simultaneous LV pacing to the RV pacing mode at either the LV apex (CRTa) or free wall (CRTf). A new index of synchrony was developed via pair-wise cross-correlation analysis of tissue Doppler radial strain from six midmyocardial cross-sectional regions, with a value of 15 indicating perfect synchrony. Compared with RA pacing, RV pacing significantly decreased radial synchrony (11.1 +/- 0.8 vs. 4.8 +/- 1.2, P < 0.01) and global LV performance (cardiac output: 2.0 +/- 0.3 vs. 1.4 +/- 0.1 l/min and stroke work: 137 +/- 22 vs. 60 +/- 14 mJ, P < 0.05). Although both CRTa and CRTf significantly improved radial synchrony, only CRTa markedly improved global function (cardiac output: 2.1 +/- 0.2 l/min and stroke work: 113 +/- 13 mJ, P < 0.01 vs. RV pacing). Furthermore, CRTa decreased LV end-systolic volume compared with RV pacing without any change in LV end-systolic pressure, indicating an augmented global LV contractile state. Thus, LV apical pacing appears to be a superior pacing site in the context of CRT. The dissociation between changes in synchrony and global LV performance with CRTf suggests that regional analysis from a single plane may not be sufficient to adequately characterize contraction synchrony.