The purpose of this paper is to present data from dynamic belt loading tests on the thorax of human cadavers where the exact timing of all rib fractures is known. In order to generate non-censored rib fracture data, a total of 47 strain gages were placed throughout the thorax of two human cadavers (1 male, 1 female). In order to simulate thoracic loading from a severe car crash, a table-top belt loading device was developed that utilizes a servo-hydraulic test machine to apply a dynamic input. The belt load pulse was configured to result in 40% chest compression through a 150 ms load and unload cycle. Potentiometers and accelerometers measured the chest compression and acceleration at three locations, load cells in line with the belt provided belt loads, and load cells on the posterior side of the thorax measured the reaction loads. The time histories of each strain gage were analyzed to determine the time of fracture which could then be compared directly to the reaction loads and chest displacements at that exact time, thereby creating a non-censored data set. In both cadavers, all fractures (20 for female and 12 for male) occurred within the first 35% compression of the thorax. As a general trend, the first series of fractures were on the left side of the thorax where the belt passed over the abdomen. The peak strain at failure ranged from 1.1 % to 2.5 %. By utilizing this technique, the exact timing of an injury level can be characterized relative to the mechanical parameters. For example, using rib fractures as the parameter for AIS scores in the female test, it is shown that AIS 1 injury occurs at a chest compression of 21%, AIS 2 at 22%, AIS 3 at 24 %, and AIS 4 at 34%.