The increased utilization of three-point restraint systems has greatly reduced the incidence of spinal injuries in motor vehicle accidents. Nevertheless, several studies which rely upon the National Automotive Sampling System (NASS) have documented lower thoracic and upper lumbar fractures in restrained occupants involved in frontal collisions of moderate severities. Although it has been postulated that the injury mechanism may be related to the occupant being out-of-position or sitting in an unusual posture, conclusions with regard to the precise mechanism of injury are difficult due to the lack of information contained in the NASS database. In addition, previous studies have not reported statistical significance of these injuries. In this study, we combined statistical analysis of frontal collisions in the NASS database with the analysis of data acquired from sled and crash tests, which utilized anthropomorphic test devices (ATDs), in order to evaluate the incidence and potential injury mechanisms underlying thoracic and lumbar spine fractures in moderate frontal impacts. In the first portion of the study, we performed a statistical analysis of the NASS database to estimate the incidence rate of spinal fracture. This was complemented with measurements and analysis of lumbar spine load data derived from frontal sled and crash tests. Analysis of the NASS database demonstrated that thoracolumbar spinal injuries are rare when an occupant is restrained by a lap and shoulder belt, and are often accompanied by abdominal injury. The spinal loads measured during frontal impacts with restrained and nominally positioned ATDs were found to be well below injury thresholds. Our results also suggest that the potential for isolated fracture is increased when the geometry of occupant-to-restraint interaction is compromised, as occurs when an occupant submarines the lap belt