Lack of information from vehicle-to-child pedestrian impacts provides considerable challenges when developing vehicle countermeasures for the pediatric population. Crash reconstructions of real-world incidents provide useful information about the vehicle damage and injury outcome but do not permit definitive and quantitative measures of the impact severity given the high level of uncertainty in the initial conditions of the pedestrian and the vehicle prior the impact. This paper develops an advanced methodology for reconstructing child pedestrian–vehicle impacts that combines the crash data with multi-body simulations and optimization techniques for identifying the pedestrian posture and vehicle speed prior to impact. For the child pedestrian posture, a continuous sequence of the running gait was developed based on the literature data and simulations. Using vehicle damage information from an actual child pedestrian crash, an objective function was developed that minimized the difference between vehicle and pedestrian contact points for the simulated child postures, pedestrian, and vehicle speeds. Simulated annealing and genetic optimization algorithms were used to identify sets of potential solutions for the pedestrian and vehicle initial conditions. Local minimums were observed for several response surfaces of the objective function which shows the non-convex nature of the crash reconstruction optimization problem with the chosen objective function. Based on the results of the real-world reconstruction, this study indicates that numerical simulations coupled with heuristic optimization algorithms can be used to reconstruct child pedestrian and vehicle pre-impact conditions.