Thoracic injuries are a major cause of mortality in frontal collisions, especially for elderly female front seat passengers. Car occupant individual characteristics are known to influence human vulnerability in crashes. In‐vivo experiments allow more freedom in the selection of the subjects than PMHS tests and could therefore contribute to quantify age, gender and geometry influence on thoracic response. In this goal, a noninjurious low deceleration protocol was defined for which two public crash simulators were tested and Ifsttar sled was modified. The results show that the public simulators were not suitable for the current study. Ifsttar sled pulse was finally chosen between the public simulators severities. From the sled tests tri‐dimensional sternal deflection was measured on Hybrid III 50th and 5th percentiles using stereovision and preliminary thorax force‐deflection curves were obtained on two volunteers. The results from volunteers showed a good intrasubject repeatability and inter‐subject response differences. The experimental protocol was accepted by the French Ethical Committee and will be applied on 40 volunteers.