As part of an effort to develop a suspension-culture perfusion-based process with high flow rate without the fouling and antibody retention inherent to filter-based cell-separation devices, we have evaluated and contributed to the development of the Centritech Lab centrifuge for the perfusion culture of hybridoma cells in protein-free medium. Culture start-ups showed that cell growth and monoclonal-antibody (MAb) production rates were similar in both a spinner flask and continuous centrifugation coupled to a bioreactor. The centrifuge efficiently separated viable cells from dead ones. Viable-cell recoveries were never below 98%, whereas dead-cell recoveries were usually around 80%. The cell content of the centrifuge supernatant and concentrate was strongly determined by the total amount of cells, viable and dead, in the culture broth, but an influence of the centrifugation parameters (feed rate, times of separation and discharge, and rotor speed) was observed. This understanding of the separation process inside the centrifuge is important and may apply to other similar devices. Monoclonal antibodies were not retained in the bioreactor during centrifugation perfusion. However, whereas similar growth rates were obtained in perfusion cultures using either continuous centrifugation or filtration, MAb concentrations were 35% lower in the former case. Utilization of the centrifuge in an intermittent fashion decreased the daily cell residence time outside the bioreactor, the daily pelleted-cell residence time in the centrifuge, and the frequency of cell passage to the centrifuge. This led to higher viable-cell numbers in the bioreactor and an accompanying increase in MAb concentrations, 225-250 mg of IgM L-1, equal to the performance of filter-based perfusion systems with the same cell line. It was hypothesized that having cells periodically packed at the bottom of the centrifuge insert (up to 800 x 10(6) cells mL-1) is deleterious to the culture by exposing the pelleted cells to prolonged nutrient limitations.