The aim of this paper is to evaluate and compare several methods allowing the reconstruction of real accidents involving pedestrians. These different methods have various levels of complications and are commonly used in primary or secondary safety research. They can be classified into three categories corresponding to their levels of complications. The first class concerns ”simple” methodologies based on an analytical or semi-analytical approach (”hand-calculi”) such as Searle’s model, Fall and Slide model, equations proposed by Rau et al., Simms et al., etc. The second one is more complicated and considers for example the pedestrian as a single segment as described by Wood. Finally, the last class contains the most complicated approaches and is based on three-dimensionalmultibody models. Concerning this third class, this work has been based on the PC-Crash® and Madymo® softwares. We have tested all of these methods for one of the most usual real car-to-pedestrian accident configurations:​ frontal collision with pedestrian wrap trajectory. Data issuing from two real cases have been used. They have been provided by an indepth multidisciplinary accident investigation (psychology, technical, medical). Reconstructions are thus based on driver and witness statements, on accurate information relating to material evidence (e.g. skid marks, car damage, pedestrian injuries, throw distance) and parameters fitted to vehicle and pedestrian (e.g. vehicle shape, pedestrian anthropometry, etc). Results have been compared in terms of quality of the reconstruction balanced by the limitation of the different methods. Evaluated elements are in particular the speed of the vehicle, the final position of the pedestrian, his kinematics, the impact points on the car and injuries (when the method allowed it). In parallel, methodologies have also been compared qualitatively by establishing the necessary means to apply them. In this way, the potentiality of the methods, their requirements (necessary input data, into operation bringing time, computer time) have been evaluated and reported in a general matrix. It allows us to summarize advantages and disadvantages of the different methods.