One of the most interesting and challenging applications of the engineering is the solution of the inverse problems, or in other words, those problems which the answer, or the consequence is known, but the cause is not, or the conditions that took such result are the unknowns, mainly when the problem involves a great number of variables and parameters. In this kind of problems we find the scientific approach of the ground vehicle accident reconstruction. Another theme directly related it is the collisions analysis, as in the context of an accident reconstruction, or in the context of the vehicle crashworthiness, associated to it structural integrity and their occupants' passive safety. In this paper is discussed the application of the genetic algorithms, for the treatment of the inverse problem in collisions of ground vehicles, using models of rigid or flexible vehicles. We are studying how the optimization algorithms can supply the group of variables and parameters that more probably would take to the final condition, inside of the imposed restrictions, starting from the vehicles position after a collision. Also using the same optimization procedures, and considering restrictions associated to the problem, we are evaluating the possibility of how to determine the rigidity and plasticity characteristics of the vehicles structures, starting from it deformed condition after a collision. This procedure can also supply information about the impact severity. The preliminary results reached are presented. In spite of being a natural way for the treatment of this kind of problem, there are still few authors that adopt the optimization techniques for it solution and interpretation. This is another effort in the sense of gives a scientific approach to the accident reconstruction and collisions analysis problems, protected and developed by the PUC-Rio Vehicular Systems Group.