Multisite ventricular pacing acutely improves the hemodynamic status in heart failure, though longer-term observations require invasive procedures. The hemodynamics of multisite ventricular pacing were assessed by echocardiography and peak endocardial acceleration (PEA) measured by a pacemaker sensor. PEA variations are highly correlated with those of dP/dt. Thirteen end-stage heart failure patients (left ventricular ejection fraction < 0.30) with a QRS > or = 140 ms received a DDD PEA sensor-driven pacemaker allowing right (RV), left (LV) and biventricular (BV) pacing. Ten days after implantation, standard echocardiographic parameters and variations in PEA were measured after 20 minutes at each pacing mode. The aortic systolic preejection time interval was statistically comparable between RV and LV pacing (218 +/- 24 vs 219 +/- 34 ms; P = NS), and significantly shorter with BV pacing (198 +/- 27 ms; P = 0.013). Aortic ejection duration was nonsignificantly shorter during BV pacing than during LV pacing (-.061, P = 0.09). The aortic velocity time integer increased during LV pacing versus RV pacing (+21%, P < 0.05) and during BV pacing versus RV pacing (+37%, P = 0.05). As a result, the values of the PEA variations over a 15-minute period were significantly greater during LV pacing and BV pacing versus RV pacing (+43%, P < 0.05, and +38%, P = 0.05, respectively) and were statistically comparable between BV pacing and LV pacing (9% for LV pacing, P = NS). During various ventricular pacing configurations, PEA measurements were consistent with echocardiographic data, showing comparable hemodynamic effects of BV and LV pacing. The PEA sensor is a promising tool for long-term hemodynamic monitoring and serial evaluation of the effects of multisite ventricular pacing in heart failure patients.