Two series of nine frontal sled tests were conducted to evaluate the behavior of the Hybrid III and Thor-α dummies. The first series was conducted at 50 kph with airbag and 4 kN force-limited shoulder belt and the second series at 30 kph and only a 4 kN force-limited shoulder belt. In each series, three replicate tests were conducted with each dummy and compared with three PMHS. The data provided by the same instrumentation located at the same position were compared to assess the biofidelity of both dummies. The results were mass scaled in order to account for the differences between the anthropometry of the cadaver. The good test-to-test repeatability for each dummy permitted to compare the mean value of each recorded parameter. Based on the cadaver response, the results show that the Thor-α provides responses that are more similar to those of PMHS than the Hybrid III. The flexible joints in the thoracic spine, the sternum design and the more humanlike ribcage give more similar accelerations than the Hybrid III as compared to those of the PMHS. Nevertheless, some parts have to be improved in order to better follow the behavior of the human subject. The head-neck complex, the chest, the shoulder and the pelvis of the Thor-α have a more humanlike behavior but some differences remain. The distribution of the deceleration between the components is sometimes different compared to those of the cadaver, even if the resultants are similar. The dummies and most particularly the Hybrid III are less sensitive to the change in restraint systems and tests conditions than a cadaver.