Mouse–mouse hybridomas (15 μm mean diameter) were cultivated in a simulated perfusion reactor with spin filter and external recirculation of the medium. Proteins at high concentrations, such as 10% foetal calf serum (FCS), were found to be not responsible by themselves for fouling, even at high recirculation rates. Stainless steel (10 μm pores) in contrast to polyamide (11 μm proes) led to a great accumulation of dead cells and nucleic acids on the screen, finally leading to fouling, as shown by biochemical and microscopic examinations. It is suggested that the high surface charge density of metals compared to polyamide is responsible for attachment of various residues. Stainless steel should rather be replaced by a resistant and nontoxic synthetic material, such as polyamide 66 which was successfully used. FCS should be avoided, since it seems to increased the fouling phenomenon. Moreover, the pore size of the screen should be carefully defined according to the wide size distribution of living and dead cells of the line used (33% of variation of the mean size in our case) as well as fragments. The purpose of the screen being to get rid of fragments and small dead cells, and not to wash too many new small cells, a good retention was achieved here by a 10-μm opening.