This paper addresses the multi-variable optimisation of bioreactor performance. The central idea here is to use a time series model, which is identified on-line, to evaluate the steady-state performance of the bioreactor. A simple algorithm has been developed around this theme and analysed in Kambhampati et al (1989). A fermentation process model has been developed to provide a test system for the optimisation scheme. The model takes into account the effect of temperature and dilution rate on the biomass production. For this process a multi-variable, model is identified on-line. It has been shown that the model provided optimal results.