Yes / A methodical 'design of experiment' approach is used to assess the effect of key control parameters on residual wall thickness (RWT) in Gas Assisted Injection Moulding. An empirical model is produced from which the experimental RWT can be determined at any interpolated point. This model includes only those terms with proven statistical significance. The 'true' thermal boundary conditions are determined for a 1-D approximation to the system, this is sufficient to determine the error in a simulation method that enforces coolant temperature as the mould boundary condition for temperature. It is shown that errors in heat-flux and wall temperature are small. A 3-D finite element, pseudo-concentration implementation is presented, with a novel method for simulation of internal gas injection. The simulation is shown to give good agreement with the experimental rate of growth of wall thickness as gas delay is increased; good qualitative agreement is shown for other control parameters.