This paper is part of a four year study to systematically define side impact injury in terms of the kinetic response of a suitable anthropomorphic dummy. Last year a paper was presented at the Experimental Safety Vehicle Conference in Germany which analyzed side impact dummy response and injury prediction based on cadaver data generated by the Highway Safety Research Institute. These subjects were generally older than those discussed in the current paper. This paper includes data from a number of University of Heidelberg cadaver sled tests-including padding tests which we recently found to be (1) critical for a definitive analysis and (2) previously not available.Two advanced dummies, whose design specifications are based upon biomechanical data, are currently being evaluated by the biomechanical community. The two dummies are (1) a Side Impact Dummy (SID) designed by the Highway Safety Research Institute (HSRI) and (2) the Association Peugeot-Renault (APR) dummy from France. The performance of these two candidate dummy designs is compared by a variety of techniques in seven identical tests using cadavers. These tests are chosen for use in evaluating biofidelity, repeatability, and dummy/padding interaction.Among these techniques is a cumulative variance analysis-similar to a root-mean-square analysis-of the acceleration signal for the seven unique sled/pendulum tests where acceleration response data exists for APR dummy, SID, and cadavers. This cumulative variance approach allows an objective comparison of the response of each dummy design with respect to the cadaver data.