Evaluation of safety benefits is an essential task during design and development of pedestrian protection systems. Comparative evaluation of different safety concepts is facilitated by a common metric taking into account the expected human benefits. Translation of physical characteristics of a collision, such as impact speed, into human benefits requires reliable and preferably evidence-based injury models. To this end, the dependence of injury severity of body regions on explanatory factors is quantified here using the US Pedestrian Crash Data Study (PCDS) for pedestrians in frontal vehicle collisions. The explanatory and causal factors include vehicle component characteristics, physiological and biomechanical variables, and crash parameters. Severe to serious injuries most often involve the head, thorax and lower extremities. In terms of causing components;​ severe head and thorax injuries occur mainly on the windshield and hood region;​ serious lower extremity injuries usually occur on the front bumper. In order to formulate a common metric for evaluating the effect sizes of distinct causal factors, multivariate models of injury severity are obtained by binary logistic regression. Impact speed is clearly the most important injury severity predictor for every body region, although biomechanical and physiological variables as well as geometrical characteristics of the vehicle are also significant in multivariate analysis, even controlling for speed. The injury-metric approach is illustrated in a comparison of hypothetical active and passive safety measures. The relative risk reduction to various body regions by this reduction is formulated as a benchmark for comparison of benefits from proposed structural changes to individual vehicle components.