Dimensional control of extrudates has been the subject of a number of publications. Good correlation between die swell and single point dynamic data, such as Tan delta, has been found for small numbers of compounds but fails when applied to numbers of large batch in production. Simulation based on multi-mode rheological models have been used with success, but require precise knowledge of both dynamic properties and steady shear viscosity versus shear rate. Results presented in this paper show good correlation with die swell and all process variables and compound dynamic properties. Process variables were: extruder barrel temperature, screw speed and temperature, extruder pressure and head temperature, compound dynamic properties were G', G" and Tan delta versus frequency and strain. Due to the large number of variables, correlation has been obtained using Neural Network modelling. This provides the opportunity to control die swell by variation of the process conditions before, rather than after extrusion of the compound, as with the most common feed back loops in operation today.