This study aims to investigate the relationship between the number of rib fractures and the thoracic deflection in side impact, and in particular its variability with respect to various loading configurations. The relevance of thoracic deflection as an injury criterion depends on the existence or not of this variability. Few studies were dedicated to this issue in the literature. First, a validation database was established, which covers different impact directions (frontal, lateral and oblique), different loading types (impactor, belt and airbag), and different injury levels (from the absence of, to presence of numerous ribs fractured). The HUMOS human body model was then modified and validated versus the database. Besides the typical validation in terms of global response, particular attention was paid to validate the model with respect to the ribcage strain profile, the occurrence of rib fractures and their locations. Some key features relative to the thorax modeling (such as the connection between the ribcage and the surrounding tissues) were also investigated. Using this model, the relationship between the rib fracture and thoracic deflection was studied versus loading type. It was demonstrated that this relationship does not change significantly from one loading type to another, supporting the thoracic deflection as a relevant indicator of ribcage injury level. In addition, the relationship between the thoracic response and the test severity (such as impact velocity or distance between airbag and subject), was also established for each kind of loading. These relationships, obtained over a unique subject, are free of uncertainty relative to cadaver scattering, measurement and autopsy. They provide a consistent characterization of thoracic injuries under various test conditions and therefore, can be used to guide cadaver test protocol development and result analysis.