This paper describes a study to investigate the application of state estimation and parameter adaptive control in fed‐batch fermentation for penicillin production. A model of the penicillin fermentation process is outlined and shown to be capable of matching industrial data. The penicillin mould biomass is controlled to a reference trajectory by applying self‐tuning control to an estimate of biomass, obtained from measurements of carbon dioxide production rate and fermenter volume and by using the fermentation model with an extended Kalman filter. The manipulated variable, sugar feed to the fermenter, is calculated by an adaptive algorithm. Simulation results compare the performance of self‐tuning control with that obtained using digital Proportional plus Integral control.