Around one third of serious injuries sustained by belted, non-ejected occupants in pure rollover crashes occur to the spine. Dynamic rollover crash test methodologies have been established in Australia and the United States, with the aims of understanding injury potential in rollovers and establishing the basis of an occupant rollover protection crashworthiness test protocol that could be adopted by consumer new car assessment programmes and government regulators internationally. However, for any proposed test protocol to be effective in reducing the high trauma burden resulting from rollover crashes, appropriate anthropomorphic devices that replicate real-world injury mechanisms and biomechanical loads are required. To date, consensus regarding the combination of anthropomorphic device and neck injury criteria for rollover crash tests has not been reached. The aim of the present study is to provide new information pertaining to the nature and mechanisms of spine injury in pure rollover crashes, and to assist in the assessment of spine injury potential in rollover crash tests. Real-world spine injury cases that resulted from pure rollover crashes in the United States between 2000 and 2009 are identified, and compared with cadaver experiments under vertical load by other authors. The analysis is restricted to contained, restrained occupants that were injured from contact with the vehicle roof structure during a pure rollover, and the role of roof intrusion in creating potential for spine injury is assessed. Recommendations for assessing the potential for spine injury in rollover occupant protection crash test protocols are made.