Experimental data of polymer behaviour at high strain rates in the literature, mostly obtained with a split Hopkinson pressure bar (SHPB), shows a dependence on the thickness of the specimen. This paper investigates this so-called thickness effect and tries to clarify the doubts on the equilibrium assumption in SHPB tests on polymers by numerical simulations. Fictitious specimens of different thickness with rate sensitive polymer-like constitutive model (Eyring's model) are used to simulate basic data of SHPB test. The comparison between the given behaviours and those derived from the simulated tests provides an objective appreciation of the quality of tests on polymers (with thin or thick specimen). Results of the simulation prove that the equilibrium assumptions can still be safely used in the case of test on polymers, and indicate that the thickness effect is mainly due to the radial inertia and friction considerations.