Rollover crashes are a serious public health problem in United States, with one third of traffic fatalities occurring in crashes where rollover occurred. While it has been shown that occupant kinematics affect the injury risk in rollover crashes, no anthropomorphic test device (ATD) has yet demonstrated kinematic biofidelity in rollover crashes. Therefore, the primary goal of this study was to assess the kinematic response biofidelity of six ATDs (Hybrid III, Hybrid III Pedestrian, Hybrid III with Pedestrian Pelvis, WorldSID, Polar II and THOR) by comparing them to post mortem human surrogate (PMHS) kinematic response targets published concurrently; and the secondary goal was to evaluate and compare the kinematic response differences among these ATDs. Trajectories (head, T1, T4, T10, L1 and sacrum), spinal segment (head-to-T1, T1-to-T4, T4-T10, T10-L1, and L1-to-sacrum) rotations relative to the rollover buck, and spinal segment extension/compression were calculated from the collected kinematics data from an optical motion tracking system. Response differences among the ATDs were observed mainly due to the different lateral bending stiffness of the spine from their varied architecture, while the additional thoracic joint in Polar II and THOR did not seem to provide more flexion/extension compliance than the other ATDs. In addition, the ATD response data were compared to PMHS response corridors developed from similar tests for assessing ATD biofidelity. All of the ATDs, generally, drifted outboard and upward during the tests similar to the PMHS. However, accompanied with this upward and outward motion, the ATD head and upper torso pitched forward (~10 degrees) while the PMHS' head and upper torso pitching rearward (~10 to ~15 degrees), due to the absence of flexion/extension compliance in the ATD spine. The differences in these pitch motions resulted in a difference of 130 mm to 160 mm in the longitudinal position of the head at 195 degrees of roll angle. Finally, substantially less lateral spinal bending was also observed in the ATDs compared to the PMHS. The results of the current study suggests there is greater upper spine flexion/extension, and lateral bending stiffness in all of the ATDs in comparison to the PMHS, and provided information for improvement of ATD biofidelity in future for rollover crashes.