In order to investigate the difference in microbubble generation between roller and centrifugal pumps, this quantitative bench study was conducted. Using a mock circuit and fresh bovine blood, a roller pump or a centrifugal pump was run with 51/min of flow and 350 mm Hg of total pressure head. Microbubbles were produced by inflow tube occlusion (3 or 5s). Blood temperature was maintained at 25° and 36°C. Using a Doppler microbubble detector, the maximum diameter of the bubbles was monitored every 0.1s for 60s (total, 600 samples) at the prearterial filter (pump outflow side) and the postarterial filter sites. The number of microbubbles larger than 32μm were counted. The roller pump generated significantly more large microbubbles than the centrifugal pump. Substantial numbers of large microbubbles were encountered at the postfilter site, particularly when the roller pump was used. The centrifugal pump does not create excessive negative pressure by inflow occlusion; microbubble generation was less than that of the roller pump. Furthermore, since the centrifugal pump has a pressure discrepancy in the pump, small bubbles tend to stay around the center of the pump. This study confirmed that the centrifugal pump generated fewer microbubbles than the roller pump during inflow obstruction. Therefore, the centrifugal pump is safer in terms of microbubble generation. Additional care should be taken to prevent air emboli while using the roller pump because substantial numbers of microbubbles may go through the arterial filter.