For many years calorimetry has been recognized as a powerful and universal tool for monitoring chemical and biological processes. A laboratory-scale reaction calorimeter (RCl, Mettler-Toledo), initially developed for chemical reaction studies with a sensitivity of 100-150 mW/l, has been improved to enable the monitoring of very low heat production rates (< 10 mW/l). A major limitation to successful process control, has been the inability to achieve real-time quantitative calorimetry. This is in part due to the operating principle of the RCl, in which the measured heat signal is calculated from the temperature difference between the reaction mass and the jacket oil and the heat transfer coefficient (UA). The latter frequently varies during a reaction, particularly a bioreaction, due to changes in volume, viscosity and cell density, and is difficult to determine accurately during the process. In the present study, this problem has been solved by a technical modification to the reactor vessel of the RCl. This involves forcing the heat transfer to occur through a well defined and constant area through the creation of a large resistance to heat transfer in the upper part of the reactor vessel. This was achieved by creating an air gap between the reactor contents and the reactor wall through the insertion of a PTFE sleeve. Control experiments undertaken with this modified system, in the absence of any reaction, showed that UA remained constant for volume changes as large as 50% of the working volume. Similarly, a simulated fed-batch experiment with monitoring of the stirring power, showed that the baseline heat signal could be accurately and quantitatively corrected for large dynamic variations of the volume. Using monitoring of the oxygen uptake rate as a reference, this modified system was validated by application to fed-batch cultures of Bacillus sphaericus 1593M. This strictly aerobic bacterium produces parasporal insecticidal crystal proteins which are toxic to mosquito larvae. In these fed-batch cultures, the nutrient feed was controlled using measurement of the metabolic heat release, since the latter is proportional to the substrate uptake rates for a given metabolic state. A culture, composed of two repetitive fed-batch cycles followed by a batch cycle, demonstrated that real-time and quantitative signals could be obtained, even for highly dynamic processes in which the volume and agitation rate may vary significantly and where quick repetitive inoculations can be made. The result of this work is a modified RCl (or Bio-RCl) which is as easy to use as any conventional bioreactor yet has the unique feature of being able to provide an accurate measurement of the energy dissipated as heat in chemical or biological processes, over a wide range of operating conditions.