Blind prediction and post-experimental studies of the flexural cyclic response of a U-shaped core wall, tested at UCLouvain, Belgium, are presented. The tested specimen (referred to as UW1) was modelled using a relatively simple 3D force–displacement-based version of the beam–column multiple-vertical-line-element (MVLEM-FD), developed at the University of Ljubljana (UL), using standard element parameters. The blind prediction accurately identified critical failure points within the specimen. However, the predicted type of failure, attributed to the rupture of longitudinal bars in the boundary region of flanges was somewhat different from that observed in the experiment. These bars were indeed ruptured, but their rupture was preceded by their buckling and the spalling of the surrounding concrete core. Moreover, the predicted force–displacement hysteretic loop exhibited a slight shift compared to experimental observations. Post-experimental studies revealed that this discrepancy was primarily due to the modelling of the loading at the top of the wall rather than the modelling of the wall itself. Specifically, an additional bending moment induced by the eccentricity of the vertical load applied using three actuators, which was overlooked in the blind prediction. Upon correcting the loading, the agreement between the numerical studies and experiment was considerably improved in terms of both the global and local response quantities. Additionally, minor improvements in modelling the boundary regions of the wall’s flanges, considering the relatively large unconfined concrete cover, were introduced, leading to enhanced estimations of near-collapse response.