Measles virus (MV) has a natural affinity for cancer cells and oncolytic MV preparations have therefore been investigated in several clinical trials as a potential treatment for cancer. The main bottleneck in the administration of oncolytic MV to cancer patients is the production process, because very large doses of virus particles are required for each treatment. Here, we investigated the productivity of different host cells and found that a high infection efficiency did not necessarily result in high virus yields because virus release is also dependent on the host cell. As well as producing large numbers of active MV particles, host cells must perform well in dynamic cultivation systems. In screening experiments, the highest productivity was achieved by Vero and BJAB cells, but only the Vero cells maintained their high virus productivity when transferred to a stirred tank reactor. We used dielectric spectroscopy as an online monitoring system to control the infection and harvest times, which are known to be critical process parameters. The precise control of these parameters allowed us to achieve higher virus titers with Vero cells in a stirred tank reactor than in a static cultivation system based on T-flasks, with maximum titers of up to 10(11) TCID50 ml(-1). (C) 2017 American Institute of Chemical Engineers