Large-scale propagation of replication-defective adenovirus vectors has not been well studied to date. One of the challenges for efficient propagation at large scale is to overcome the sensitivity of virus infected cells to gas sparging required for oxygenation and CO2 removal. In our initial experiments, it was observed that productivity of an adenovirus vector was significantly reduced under sparging conditions as compared to nonsparged, i.e., surface-aerated controls in serum-free cultures. Investigations led to the identification of a buffer containing surfactant (Polysorbate-80, PS-80) that was included in the virus seed stock formulation and introduced through virus infection into the culture at a very low concentration as the cause of the reduced virus productivity. This finding was not obvious and trivial, as neither uninfected sparged nor infected nonsparged PER.C6(TM) cells in serum-free cultures were affected by the buffer at such a low PS-80 concentration of 0.00025% (v/v), which is a common component of serum-free cell culture media. These results strongly suggest that virus-infected cells behave very differently from uninfected cells under sparging conditions. To mitigate the deleterious effects of sparging, the virus seed stock was prepared in the absence of the buffer containing PS-80. At the same time, the concentration of Pluronic-F68 (PF-68) in the serum-free medium was increased to 1 g/L, at which cell growth and metabolism were unaffected, even though this measure alone did not result in virus productivity improvement. Only by implementing the two measures together was virus productivity loss completely eliminated under sparging conditions. After demonstration of the process robustness in 2-L bioreactors, this adenovirus propagation process was successfully scaled up to 250 L in a 300-L bioreactor under the worst-case sparging conditions projected for 10,000-L scale. (C) 2003 Wiley Periodicals, Inc.