The mathematical model for melt spinning of Doufas et al. [Doufas, A. K. et al., J. Rheol. 43, 85-109(1999); J. Non-Newtonian Fluid. Mech. 92, 27-66 (2000); 92, 81-103 (2000)] coupling the polymer microstructure (molecular orientation, chain extension, and crystallinity) with the macroscopic velocity/stress and temperature fields is tested against low-and high-speed spinline experimental data of PET melts. The model includes the combined effects of flow-induced crystallization (FIC), viscoelasticity. filament cooling, air drag, inertia, surface tension, and gravity and simulates melt spinning from the spinneret down to the take-up roll device (below the freeze point). As is the case with nylon systems, model fits and predictions are shown to be in very good quantitative agreement with spinline data fur the fiber velocity, diameter, and temperature fields at both low- and high-speed conditions, and, with flow birefringence data available for high speeds. Our model captures the necking phenomenon for PET quantitatively and the associated extensional softening which is shown to be related to nonlinear viscoelastic effects and not to the release of latent heat of crystallization. Although crystallization is quite slow under low-speed spinning conditions, the model captures the occurrence of the freeze point naturally, and is thus a significant improvement over existing melt spinning models that enforce the freeze point at the glass transition temperature. In this article we demonstrate the robustness of our microstructural FIC model to melt spinning of quite slow crystallizers in the quiescent state, while the robustness for faster crystallizers was shown previously [Doufas, A. K. er al., J. Non-Newtonian Fluid. Mech. 92, 27-66 (2000); 92, 81-103 (2000)]. (C) 2001 The Society of Rheology.