It is acknowledged that to significantly improve biochemical system performance control should be implemented in real-time. Control algorithms, particularly modem control algorithms, require knowledge of the process dynamics and continuous measurement or detection of the system states and outputs. In many processes state variables cannot be measured on-line due, for example, to the non-availability of on-line sensors. Estimation techniques can be applied to estimate nonmeasurable state variables. Such techniques in general require the use of accurate system models. In this thesis simplified yet nonlinear models of fed-batch and batch fermentation processes are presented. Details of modelling and simulation studies carried out on a Baker’s Yeast fermentation process are included. Variations of the basic models are considered and tested using computer simulation with a view to evaluating the effect of different influences on the specific biomass growth rates. Results presented using experimental data and computer simulation results indicate the validity of a number of estimation procedures including an observer, extended Kalman filter and an iterative extended Kalman filter. Adaptive recursive least squares is applied to identify the uncertain parameters which influence the growth phases of biomass (e.g. the yeast organism).