Résumé : Studies reporting spine kinematics during cervical manipulation are usually related to continuous global head–trunk motion or discrete angular displacements for pre-positioning. To date, segmental data analysing continuous kinematics of cervical manipulation is lacking. The objective of this study was to investigate upper cervical spine (UCS) manipulation in vitro. This paper reports an inter- and intra-rater reliability analysis of kinematics during high velocity low amplitude manipulation of the UCS. Integration of kinematics into specific-subject 3D models has been processed as well for providing anatomical motion representation during thrust manipulation.Three unembalmed specimens were included in the study. Restricted dissection was realized to attach technical clusters to each bone of interest (skull, C1 to C4 and sternum). During manipulation, bone motion data was computed using an optoelectronic system. The reliability of manipulation kinematics was assessed for three experimented practitioners performing two trials of 3 repetitions on two separate days.During UCS manipulation, average global head-trunk motion ROM (±SD) were 14±5°, 35±7° and 14±-8° for lateral bending, axial rotation and flexion-extension, respectively. For regional ROM (C0-C2), amplitudes were 10±5°, 30±5° and 16±4° for the same respective motions. Concerning the reliability, mean RMS ranged from 1° to 4° and from 3° to 6° for intra- and inter-rater comparisons, respectively.The present results confirm the limited angular displacement during manipulation either for global head-trunk or for UCS motion components, especially for axial rotation. Additionally, kinematics variability was low confirming intra- and inter-practitioners consistency of UCS manipulation achievement.