In the repetitive static plate load test, a plate is loaded in two cycles where the second loading cycle provides a modulus denoted Ev2. When calculating the stiffness of a railway embankment, Young’s modulus is often assumed to be equal to Ev2 throughout the embankment. This approach, however, provides inaccurate results, mainly due to soil nonlinearity and the influence of confinement stress. Currently, there exists no method to account for these aspects to derive reliable deformation properties of embankments. Occasionally, correction factors are applied to Ev2, resulting in crude estimations. In this study, plate load tests were simulated in PLAXIS 2D using the Hardening Soil Model and calibrated against four field tests, conducted on crushed rock-fill sub-ballast. The calibrated soil properties were applied in finite element simulations of railway embankments with ballastless slab-track systems. Based on the results of finite element analyses, a semi-empirical approach is proposed, which considers confinement stress through a hyperbolic stress–strain relationship. Soil properties for compacted rock-fill with particle grading 0–150 mm were assumed through the results of the calibrated finite element analyses and the method was verified against 43 plate load tests. This semi-empirical method is more accurate than assuming a constant Young’s modulus, while maintaining simplicity and ease of use.