The efficiency of left ventricular assist devices (LVADs) depends on the capacity of the inflow cannula to drain blood into the pump. Left atrial (LA) and left ventricular (LV) sites were compared in an animal model mimicking different hemodynamic conditions. Three calves (56.3+/-5.0 kg) were equipped with a Thoratec LVAD. A regular cardiopulmonary bypass (CPB) circuit was used as a right ventricular assist device (RVAD) (jugular vein/pulmonary artery), and preload conditions were adjusted by storage (or perfusion) of blood into (or from) the venous reservoir. LA and LV drainage, tested separately or simultaneously, was measured by its effect on the LVAD's performance. The LVAD was used alone on a beating heart or together with the RVAD (biVAD) on a beating and on a fibrillating heart. Increasing the central venous pressure (CVP) highlighted the differences between the LA and LV cannulation sites when the LVAD was tested either alone or together with the RVAD (biVAD) on a beating heart. Drainage through the LA or the LV was similar when CVP was set at 8 mm Hg, and increasing CVP to 14 mm Hg allowed for better drainage through the LV cannula. In contrast, after induction of fibrillation to mimic extreme heart failure, the drainage was better through the LA cannula. Using both LA and LV cannulae simultaneously did not improve the LVAD output in any of the conditions tested. LV cannulation provides better blood drainage when used on a normal beating heart and, therefore, allows for increased LVAD performance. However, in severe heart failure, blood drainage through the LV cannula decreases and the LA cannulation site is superior.