This study compares head impact dynamics between post mortem human surrogates (PMHS) and the Polar-II pedestrian crash dummy in vehicle-pedestrian impacts with a small sedan and a large SUV. A total of fifteen (8 sedan, 7 SUV) full-scale vehicle pedestrian impact tests were performed at 40 km/h. For each vehicle, two (SUV) or three (sedan) PMHS tests and five dummy tests were performed, with three of the dummy tests in the same configuration to show repeatability, and the other two tests utilizing slightly different configurations. Head linear and angular kinematics were captured from PMHS and dummy head instrumentation, and dummy neck forces and impact forces were calculated from the upper neck load cell data. Differences in head impact locations, timing, and kinematics between the dummy and PMHS were minimized when the dummy was positioned higher above the ground reference level to match the pelvis height of the PMHS. On average, the dummy recorded higher resultant impact forces (2930 N vs. 1862 N) in windshield impacts to the sedan than in hood impacts to the SUV, which resulted in higher HIC15 values and higher peak and averaged angular accelerations. While differences in dummy injury risk metrics both the dummy and PMHS data show that the difference in injury risk metrics predicted by the dummy can be explained by the variation in impact velocity between the sedan (14.1 ± 1.2 m/s) and the SUV (10.7 ± 2.3 m/s), the differences in injury risk predicted by the PMHS is not as clear due to confounding factors. The data and analyses presented in this study also show that neck forces during head impacts contribute a substantial and additive effect to the head impact accelerations (and thus HIC15 values) measured in the dummy, and that for the SUV, neck forces affect head accelerations more than impact forces. Despite analyzing only lateral impacts with two vehicle geometries at 40 km/h, this study provides the only comparison of PMHS and dummy pedestrian head impact kinematics data available.