Belt interaction with the dummy's chest or pelvis was investigated during simulated frontal decelerations to develop a better understanding of the mechanics of belt restraint. Hyge sled tests were conducted at acceleration levels of 6-16 g's with a Part 572 dummy forward facing on an automotive bucket seat. Dynamics were compared in similar tests where the dummy was restrained by a conventional shoulder belt or belt segments attached to a modified sternum - a steel sternum with extensions for fixed belt attachments. Tests were also conducted with a conventional lap belt or belt segments fixed to an extension of the H point. Deformation characteristics of the standard and modified thorax were determined for a lateral and superior point load or a belt yoke compression of the sternum. The pelvic structure was also compressed by a lap belt.Our evaluation of test dummy dynamics indicates the following sequence of events with a conventional shoulder belt: 1.) Slack is taken up by forward movement of the dummy. 2.) The belt slips 5.1 ± 1.1 cm over the chest at a maximum velocity of 3.8 ± 1.0 m/s -independent of deceleration level. The belt then adjusts to the chest. 3.) Restraint involves primarily planar deceleration and compression of the thorax as belt loads increase. 4.) In the final phase the chest rotates about the belt and undergoes significant lateral deformation - as much as 5 cm shift in the midsagittal line - as the head and upper extremities decelerate. The nonsymmetric deformation of the thorax provides a pocket in the leather sternum for the diagonal shoulder belt. Interestingly, maximum thoracic deformation occurs after the peak in thoracic deceleration or belt load. In tests with fixed belt attachments to the chest, belt force develops faster to a higher peak and is of a shorter duration.The following sequence of events was observed in tests with the lap belt: 1.) belt slack is taken up and the belt adjusts to the pelvis, 2.) the pelvis rotates about the belt and translates along the arc of the belt, and 3.) maximum restraint occurs as the plastic covering of the pelvis compresses up to 8 cm and the upper extremities articulate forward and decelerate. Again rigid belt attachments to the pelvic structure result in a faster rise to a higher peak belt force which is of a shorter duration. Our study provides basic new information on the mechanics of belt interaction with the dummy, which may be useful in future efforts to improve belt restraint models in occupant dynamics programs.