The strategies for control of the feed rate in high-density perfusion cultures of animal cells are limited to several simple schemes. While in an industrial environment simplicity is seen as a major advantage, the need for more elaborate closed-loop control methods that can improve process stability in long-term continuous cultures is also well understood. What has prevented the application of the advanced control strategies known from theory is the lack of reliable real-time information that can be used to close the feedback loop. Among the variables that are appropriate for direct feedback control of the perfusion rate, high priority should be given to the glucose concentration. Unlike some other environmental variables, such as dissolved oxygen and pH, it provides unambiguous information which facilitates the selection of the right feed rate. The present paper describes the application of a closed loop control scheme, known as a "glucose-stat", to the long-term cultivation of Chinese hamster ovary cells in a high-density (35-40 million cells/mL) perfusion process. The monitoring and control system worked successfully for more than 2.5 months without any signs of performance degradation. In targeting industrial application, issues such as reliability, sterility, and accuracy, are given high priority. The implementation of the glucose monitoring system, which is the main part of the control complex, is addressed in details. The performance of the perfusion culture was evaluated at four different glucose set points, providing essential information about process optimization. It became evident that the perfusion culture was operated in the so-called "high-gain" zone (where the system is highly sensitive to the dilution rate), which justifies the application of a feedback control. The on-line glucose concentration was also used by an embedded expert system which drove the process through the batch and the perfusion phase, achieving total computer control of the feed rate. In summary, the proposed glucose monitoring and control technique proved to be a reliable biotechnology tool which can be applied with confidence at an industrial scale to either microbial or mammalian cell cultures.