While rollover crashes are rare, approximately one third of vehicle occupant fatalities occur in rollover crashes. Most severe-to-fatal injuries resulting from rollover crashes occur in the head or neck region, due to head and neck interaction with the roof during the crash. While many studies have used anthropomorphic test devices (ATDs) to predict head and neck injury, the biofidelity of ATDs in rollover has not been established. This study aims to build on previous research to compare the dynamic response and injuries sustained by four post mortem human surrogates (PMHS) to those predicted by six different ATDs in full-scale rollover crash tests. Additionally, this study evaluates injuries sustained by PMHS relative to possible contributing factors including occupant kinematics, occupant anthropometry, and vehicle roof deformation. While the vehicle kinematics and roof deformation were comparable for all tests, three out of the four PMHS sustained cervical spine injury, but only the tallest specimen sustained cervical spine fracture. Neck flexion at the time of head-to-roof contact appears to have affected cervical spine injury risk in these cases. Despite the injuries sustained in the PMHS, none of the six ATDs measured forces or accelerations that exceeded injury assessment reference values (IARVs), which adds to recent literature illustrating substantial differences between ATDs and PMHS in a rollover-like scenario.