Accident reconstruction is a scientific study field that depends on analysis, research and drawing. Scientific reconstruction of related traffic accident on computer eliminates making decisions depending on initiative or experience of the expert and yields impartial decisions and evidences especially on events like matter for the courts or forensic investigations. In this study, data collected from accident scene (police reports, skid marks, deformation situation of involvements, crush depth etc.) were inserted properly into the software called "vCrash" which is able to simulate the accident scene in 2D and 3D. Then, 784 parameters, related to calculating Energy Equivalent Speed (EES) with a prediction error, were prepared according to several accidents. These parameters were also used as teaching data for the Multi-layer Feed Forward Neural Network (MFFNN) and Generalized Regression Neural Network (GRNN) models in order to predict EES values of involvements, which give idea about severity and dissipation of deformation energy corresponding to the observed vehicle residual crush, without requirement of performing simulation for probable accidents in future. Using 10-fold cross validation on the dataset, standard error of estimates (SEE) and multiple correlation coefficients (R) of both models are calculated. The GRNN-based model yields lower SEE whereas the MFFNN-based model yields higher R.