Child dummies used in certification dynamic tests have not been improved since their marketing and their approval as European regulation dummies. Their main shortcoming lies in a too high and therefore unrealistic stiffness of the torso front part. The paper addresses a study carried out in the aim of solving this problem. It includes two parts: in a first section, the changes brought to the dummy torso and intended to improve its biofidelity and to reduce stiffness drastically are described. In order to reach such an objective, the lower part of the upper torso was remodelled; the pelvis profile was redefined and the geometrical and mechanical characteristics of the foam used for the abdominal insert were changed. The results obtained using two transducers installed in the abdominal section are then presented. The measurement principle of the first transducer consists in a pressure measurement, and the principle of the second one in a load measurement. Modified dummy kinematics significantly differs from that of the standard dummy when the probability of submarining occurence is high. Globally, the results recorded for the two transducers inserted in the abdominal section are similar for all tested configurations. The signals recorded will be presented and interpreted to highlight clearly submarining occurence conditions and to better analyse injury mechanisms of the abdominal section.