This paper follows Part I that was presented at the 1999 IRCOBI Conference. A methodology, described in Part 1, has been developed that permits the reconstruction of certain incidents that occur in American football. Twenty-four cases of helmeted head impact, for which concussion was diagnosed in 9 cases, have been replicated with Hybrid III ATDs. Rigid body translational and rotational head accelerations have been measured in each case. Correlations between head injury and head kinematics have been sought. Peak translational and peak rotational acceleration and velocity, HIC and the Gadd SI, as well as the GAMBIT have all been considered. A new approach employing the maximum value of the global rate of energy dissipation has proven to provide the best correlation between concussion probability and head kinematics. This new relationship provides a basis for a new head injury criterion function, the HEAD IMPACT POWER.