The goal of this study was to quantify and examine differences between the whole‐body occupant kinematics of the Hybrid III 50th Percentile Male ATD and the Hybrid III 50th Percentile Pedestrian ATD in controlled laboratory tests aimed at simulating the airborne phase of a rollover crash. Eight pure roll tests (without ground impact) were performed where a vehicle test buck, which was pitched front‐end down by 5 degrees, was rotated at a quasi‐static, low (180 deg/s) or high (360 deg/s) roll rate (pure rotation tests). In each test, one of the two ATDs was seated in either the trailing‐ or leading‐side front‐row seat with three‐point belt restraints. ATD kinematics was measured using a 500‐Hz 26‐camera 3D motion capture system. The buck and ATD kinematics were highly repeatable in these tests. Both ATDs, generally, drifted outboard and upward as a result of the rotations applied, and the ATD torso also pitched forward, which moved the head forward and the pelvis rearward. The ATD kinematics and excursion were the combined result of centrifugal acceleration (from the buck’s rotational motion), gravity and inertia (acceleration and deceleration of the buck). It was also observed that the Hybrid III Pedestrian ATD had a smaller vertical excursion and forward pitch rotation, but larger lateral excursion compared to the Hybrid III ATD. This study established a repeatable methodology for measuring occupant kinematic response when vehicle kinematics simulated rollover crashes, which can be used to facilitate rollover occupant kinematics analysis, ATD biofidelity assessment and computational model validation.