Virtual testing of helmets using finite element (FE) analysis can be a valuable tool during product development. Still, its usefulness is limited by the quality of the constitutive model of the energy-absorbing material, usually foam. Built-in constitutive models in commercial FE software are developed for traditional linear compression loading. However, modern oblique test methods load the foam in combined compression and shear. Therefore, we aim to evaluate to what extent built-in constitutive models in commercial FE software can represent Expanded Polystyrene (EPS) foam during combined compression and shear loading (CCSL). EPS foam is tested experimentally in a newly developed test rig for CCSL (V-test). The response is compared against the simulation using three different constitutive models available in LS-DYNA (M83, M126, and M181). The models are assessed by their ability to capture the correct response, focusing on how well the continuum models can capture the phenomenological events seen in the experiments. The results show that the models perform well in compression, as expected. However, we point out limitations in the shear response and significant limitations in the unloading response, both important for oblique helmet testing. Due to these limitations, we conclude that the existing models are inadequate for accurately simulating oblique helmet impacts. There is a clear need to develop and implement new constitutive models focused on capturing CCSL including the unloading. Additionally, frictional sliding was found to substantially influence the measured response in the V-test method. Minimizing interface sliding is therefore critical for isolating the material behavior.