An on-line estimation of the specific growth rate of an aerobic fermentation batch process was conducted based on measurements of oxygen or carbon dioxide concentrations in exhaust gas, or the concentrations of dissolved oxygen in the liquid phase in the fermentor. A simple linear model used was upgraded with a mechanism for a time-varying forgetting factor to enable good tracking of estimated parameters as well during rapid and large changes in the process. Because the measuring methods used are independent of each other and can identify the process well, mutual control of individual sensors is also possible in industrial reactors. The results from the model agree well with experimentally obtained values for bacitracin fermentation in an 80-m(3) and 1.5-m(3) batch bioreactor. The model provides a good basis for robust adaptive regulation of aerobic batch processes.