The Molecular Strain Function model with Convective Constraint Release has been demonstrated by Wagner to fit elongational and shear viscosities, and First and Second Normal stress differences for a variety of polymer melts, when used with a Convective Constraint Release mechanism [J. Rheol. 45 (2001), 1387]. A modification to the CCR mechanism was shown to give more accurate representation of corner vortices in an abrupt contraction flow [JNNFM 135 (2006), 68] for both planar and axisymmetric contraction flows. It is highly desirable to assess the model against 3D flows. A primary advantage of 3D simulation in assessing a constitutive model is that, experimentally, it is very difficult to produce truly 2D data; the side walls of a finite die affect stress birefringence measurements (since this is a ‘line of sight’ cumulative measurement), and also induce significant 3D motion into the flow. The existing 2-dimensional code has been extended to fully 3-dimensional flows using 27-node ‘brick’ elements, and using a number of developments to deal with tracking and storage problems inherent in 3D time-integral solution. The 3D code is assessed against known 2-dimensional solutions to verify its accuracy; the constitutive model is then assessed against experimental data for a 4:1 contraction ratio die, which has finite width (5:3 depth to inlet height), inducing 3D effects. Stress birefringence, vortex size, and cross-sectional flow rate data at a number of flow rates are compared. The model is shown to give good accuracy against this flow.