The study reports on the results of frontal collisions with 16 cadavers and two Hybrid III dummies with impact velocities of 48 km/h to 55 km/h and a mean sled deceleration of 17 g; mounted to the sled was the front part of a passenger compartment. The cadavers were restrained in the driver position with either 3-point belts (6% and 16 % elongation) and/or air bag with knee bolster and one case was unrestrained. In most cases, both a 12-accelerometer thoracic array and 2 chest bands were employed. In some cases the acceleration at Th6 was measured. The cadavers were autopsied and the injury severity was rated according to the AIS 90. Maximum resultant Th1, Th6, and Th12 accelerations or sternum accelerations in x-direction ranged from 35g to 78g when using 3-point belts and produced injuries ranging from a few rib fractures to unstable chest wall (flail chest). The same range of acceleration values were observed when a driver air bag was used, but three of the four the subjects remained uninjured. Chest deflections of 3 cm to 7 cm were observed by using 3-point belts and/or air bags with knee bolster. The sternum fracture was the typical injury by using 3-point belt systems. The chest band contours showed the torso belt produced a high local compression with resulting higher injury risks, while the air bag appeared to load a larger area of the chest front with less injury risk. Cervical spine injuries of AIS 1 and AIS 2 were observed, independent of using 3-point belt and/or air bag systems. The 3-point belt combined with driver air bag appears to be the best solution for the protection of the head and thorax during frontal collisions. However, further modification of the torso belt's yield characteristic would appear to offer further safety improvements.