Current safety testing protocols typically evaluate performance at a single test speed, which may have undesirable side effects if vehicles are optimised to perform at that speed without consideration to performance at other speeds. One way of overcoming this problem is by using an evaluation that incorporates the distribution of speeds that would be encountered in real crashes, the relationship between test speed and test performance, and the relationship between test performance and injury risk. Such an evaluation is presented in this paper and is applied to pedestrian headform testing. The applicable distribution of pedestrian impact speeds was compiled from in-depth crash data. Values of the Head Injury Criterion across the speed distribution were imputed from a single test result, taking into account the potential for ‘bottoming out’ on harder structures beneath the hood. Two different risk functions were used: skull fracture risk and fatal head injury risk. Eight example test locations were evaluated; each had an underhood clearance such that it would perform worse at higher speeds than suggested by its original test result. When the effect of bottoming out was included in the evaluation, the calculated average injury risk was generally higher than it was if bottoming out was ignored. The average risk of fatal head injury was more affected by the inclusion of bottoming out than the average skull fracture risk. The methodology presented in this paper may be extended to other forms of impact testing, although the input functions may be more difficult to derive for more complex tests.