The National Aeronautics and Space Administration (NASA) is interested in characterizing the responses of THOR (test device for human occupant restraint) anthropometric test device (ATD) to representative loading acceleration pulse s. Test conditions were selected both for their applicability to anticipated NASA landing scenarios, and for comparison to human volunteer data previously collected by the United States Air Force (USAF).THOR impact testing was conducted in the fore-to-aft frontal (-x) and in the upward spinal (-z) directions with peak sled accelerations ranging from 8 to 12 G and rise times of 40, 70, and 100ms. Each test condition was paired with historical huma n data sets under similar test conditions that were also conducted on the Horizontal Impulse Accelerator (HIA). A correlation score was calculated for each THOR to human comparison using CORA (CORrelation and Analysis) software. A two -parameter beta distribution model fit was obtained for each dependent variable using maximum likelihood estimation.For frontal impacts, the THOR head x-acceleration peak response correlated with the human response at 8 and 10G 100ms, but not 10G 70ms. The phase lagged the human response. Head z-acceleration was not correlated. Chest x-acceleration was in phase, had a higher peak response, and was well correlated with lighter subjects (Cora = 0.8 for 46 kg vs. Cora = 0.4 for 126kg). He ad x-displacement had a leading phase. Several subjects responded with the same peak displacement, but the mean of the group was lower. The shoulder x-displacement was in phase but had higher peaks than the human response.For spinal impacts, the THOR head x-acceleration was not well correlated. Head and chest z-acceleration was in phase, but had a higher peak response. Chest z-acceleration was highly correlated with heavier subjects at lower G pulses (Cora = 0.86 for 125kg at 8G). The human response was variable in shoulder z-displacement, but the THOR was in phase and was comparable to the mean peak response. Head x- and z-displacement was in phase, but had higher peaks. Seat pan forces are well correlated, are in phase, but have a larger peak response than most subjects.