This contribution reports on the process systems engineering analysis of a food extrusion process leading towards the development and implementation of an on-line process control scheme. A versatile dynamic model of the process predicting the quality variables product gelatinisation, specific mechanical energy, specific volume and residence time has been developed and fitted to experimental data. The model is used to investigate the practicality of an adaptive inferential estimator (Guilandoust et al., 1987) for predicting infrequently measured quality variables on-line. The simulated performance of simple PI and model predictive control of these quality variables is compared.