In this paper, we propose a theoretical physical lumped parameter model for the mixing process in plasticizing self-wiping twin screw extruders using a combination of statistical techniques and mechanistic modelling. The model describes the response of the extruder to input signals of feed rate and screw speed in the time domain via residence time density distribution (RTD). It is shown that the RTD refers to the impulse response of the extruder to input signals. Parameters of the RTD function are estimated by developing predictive models of the mean residence time and mean time delay by way of mechanistic modelling, taking into consideration the flow behavior during the plasticizing process and its interaction with the screw configuration. Analyses show that the evolution of the mixing process is dependent on the feed rate (Q) and screw speed (N) combination, with the feed rate having a greater impact in changing the shape of the RTD curve. The process is optimized by determining the variance of the material distribution in the extruder. The model is presented in the form of an impulse response function (t) in the time domain suitable for the design of a controller. © 2013 IEEE.