The present investigations show the development of process models for the mixing process in an internal mixer for the prediction of compound properties, In the first approach to modelling, use was made of regression analysis. By way of an alternative, artificial neural networks were examined as a means of modelling. In the course of these studies, the mixing process was modelled both with standard compounds on a laboratory extruder and with production compounds on industrial internal mixers, employing both analysis methods. The results show that the two modelling methods permit the Mooney viscosity to be predicted with a high level of accuracy, of approximately 90%, as a function of the application in question.