To satisfy the needs of growing urban populations, mitigate the impacts of climate change, and improve the environmental conditions and quality of life in cities, there has been an increasing demand for public transport, especially effective and economically beneficial trams. As a consequence, the issue of pedestrians’ safety has become more important because of a higher risk of tram-pedestrian collisions. The article is focused on the development of a computational model that can accurately simulate such collisions, and its challenges and limitations. Experiments were conducted using various types of trams, different impact speeds, and an anthropomorphic test device (dummy). Computational models of the tram front ends were created based on CAD geometry supplied by VÚKV a.s. The finite-element mesh itself used a combination of 1D, 2D, and 3D elements. Depending on the shape of the individual components, TRIA, QUAD, HEX, PENTA, and TETRA elements were used for modeling. The results showed that the consequences of these collisions are sensitive to small differences in the posture of the dummies. Therefore, it is difficult to identify a generally more pedestrian-friendly shape for trams using the dummy-tram crash test method.