Vulnerable road users (VRUs) face high risks of injury and death from traffic accidents. The current VRU head protection program relies on a single head impact velocity and injury assessment criteria that fail to account for brain tissue strain. This limitation affects the effectiveness of simulating real-world impacts and the accuracy of head injury risk assessments. In this study, the VRU head impact boundary conditions were extracted based on the reconstruction of 40 real-world pedestrian VRU head impacts. Using the Total Human Model for Safety (THUMS) head finite element model and headform impactor, this study explored the effects of real head impact boundary conditions on head kinematics and injury under procedural test scenarios and compared these conditions with test procedure scenarios. The results indicate that the peak linear acceleration in the current test procedure scenarios is higher; however, the peak rotational velocity is significantly lower than that observed in real-world accidents. Different impact locations have significant effects on the head kinematics and injury response parameters, particularly in stiffer areas, such as the windshield edges and lower right corner, where the injury risk under regulatory conditions is higher than that in real accident cases. In contrast, the opposite is true in other windshield areas. This study suggests that future programs or virtual assessments should diversify the head impact boundaries and injury assessment criteria to consider the differences in impact locations and the effects of head rotation on brain tissue injuries. For most windshields (non-edge areas), increasing the linear velocity enhances head rotation, and rotational injury assessment criteria should therefore be introduced. For future virtual assessments, injury criteria based on brain tissue strain should be used to assess VRU head injury risk in real accidents more comprehensively and accurately.