The protection of automotive occupants against broadside collisions is of current interest due to the implementation of a new Federal standard on side impact. There is controversy as to the validity of the standard, the best criterion to assess injury to the thorax and the type and manner in which padding should be used. Although it has been shown in a series of 17 cadaveric tests that paper honeycomb can reduce thoracic injuries dramatically, there are still concerns regarding the loss of air space between the door and the occupant due to the presence of the padding and regarding the loss of protection for the thorax when there is no engagement of the shoulder with the intruding side structure of the car. This paper describes the development of a three-dimensional rigid body model to simulate cadaveric experiments carried out at Wayne State University. Model parameters were chosen to yield human-like responses at the level of the shoulder, thorax, abdomen and pelvis. The model was then used to study the effect of padding on injury parameters related to the nearside occupant when a relatively thick padding is used (up to 100 mm). It was also used to study the increase in force on the thorax when shoulder engagement is lost. Laboratory tests were conducted with full shoulder engagement but in the field most cars have a low beltline (window sill) which effectively eliminates shoulder contact if the arms are outstretched in a normal driving posture. If a sufficiently soft padding was used, the model did not predict an increase in thoracic force level or any of the injury parameters. This result contradicted those from previous models which either used a dummy as the near-side occupant or did not take into account the acceleration of the target vehicle. The model also predicted that the shoulder can provide substantial protection to the thorax and that without shoulder engagement, the force on the thorax can be approximately 45 percent higher. A door design which can lower its first velocity peak to match the second peak and which has 75 mm of soft padding is predicted by this model as being able to provide protection to the nearside occupant for moderate impacts. Additional cadaveric testing is needed to confirm these results.