For over a century, centrifugal pumps (CP) have been used in various applications, from large industrial pumps to flow pumps for aquariums. However, the use of CP as blood pumps has a rather short history. Consequently, the hydraulic performance data for a blood CP are limited. The aim of our investigation was to study the hydraulic performance and the heat generation of three commercially available CP: Bio-Medicus Bio-Pump BP80 (Medtronic), Rotaflow (Jostra Medizintechnik), and DeltaStream DP2 (MEDOS Medizintechnik AQ). The study was performed using a circuit primed with a water-glycerin mixture with a dynamic viscosity of 0.00272 pa/s. Pressure-flow curves were obtained by a stepwise stagnation of the pump outlet or inlet. The temperature changes were observed using ThermaCAM SC2000 (Flir Systems). The pumps' performance in close to clinical conditions ('operating region') was analysed in this report. The 'operating region' in the case of the BP80 is positioned around the pressure-flow curve at a pump speed of 3000 rpm. In the case of the Rotaflow, the 'operating region' was between the pump pressure-flow curves at a speed of 3000 and 4000 rpm, and the DP2 was found between 7000 and 8000 rpm. The standard deviation of mean pressure through the pump was used to characterise the stability of the pump. In experiments with outlet stagnation, the BP80 demonstrated high negative association between flow and pressure variability (r = -0.68, p < 0.001). In experiments with the DP2, this association was positive (r = 0.68, p < 0.001). All pumps demonstrated significantly higher variability of pressure in experiments with inlet stagnation in comparison to the experiments with outlet stagnation. The rise of relative temperature in the inlet of a pump was closely related to the flow rate. The heating of fluid was more pronounced in the 'zero-flow' mode, especially in experiments with inlet stagnation. In summary, (1) the 'zero-flow' regime, which is described in the manuals of some commercially-available pumps, is the use of the pump outside the allowable operating region. It is potentially dangerous and should, therefore, never be used in clinical settings. (2) Using centrifugal pumps for kinetic-assisted venous return can only be performed safely when the negative pressure at the inlet of the pump is monitored continuously. The maximum allowable negative pressure has to be defined for each type of pump, and must be based on pump performance.