The evaluation of pedestrian safety performance of vehicles required by regulations and new car assessment programs (NCAPs) have been conducted. However, the behavior of a pedestrian in an actual car-pedestrian accident is complex. In order to investigate injuries to the pedestrian lower body, the biofidelity of the lower limb and the pelvis of a pedestrian dummy called the POLAR II had been improved in past studies to develop a prototype of the next generation dummy called the POLAR III. The biofidelity of the thigh and the leg of the POLAR III prototype has been evaluated by means of 3-point bending. However, the inertial properties of these parts still needed to be adjusted to match those of a human. The biofidelity of the pelvis of the POLAR III prototype has been evaluated in lateral compression. Although the experiment using PMHSs (Post Mortem Human Subjects) was conducted in dynamic condition, the dummy tests were performed only in quasi-static condition. Therefore, this study aimed to improve and further validate the lower limb and the pelvis of the pedestrian dummy to enhance assessment capabilities of injuries to these body regions. The femur and tibia solid shafts of the POLAR III prototype were modified to improve durability of the instrumentation. The modified thigh and leg of the POLAR III prototype consist of double-layered plastics shaft covered by the flesh. These parts were subjected to latero-medial 3-point bending. The isolated pelvis of the POLAR III prototype was subjected to dynamic lateral compression. The force-deflection curves of the modified leg, thigh and pelvis of the POLAR III prototype (except for iliac reaction force of pubic loading) were compared with those of the PMHS test results from a past study, and the biofidelity of these components was evaluated using the parameter proposed by a past study.