Résumé : Objective. Determination of the three-dimensional kinematics of the head relative to the upper trunk obtained during a manipulation applied on two different cervical levels and on both sides. Design. Descriptive study performed on 14 asymptomatic volunteers. The range of motion was measured by a 3D electrogoniometer during manipulation executed by the same practitioner. Background. Spinal manipulative therapy is a common treatment approach in patients suffering from some spinal disorders. Complications exist; they are thought to be related to the force applied by the practitioner and the range of spinal motion obtained during the manipulation. Yet, little is known about cervical spine motion during manipulation. Methods. Three dimensional electrogoniometry using a 6 degree-of-freedom spatial linkage fixed between the head and the upper trunk was used to record the pattern of motion and the amplitudes obtained during a manipulation on two cervical levels (C3 and C5) and on left and right sides. On single practitioner applied the same technique to all subjects in a seated position. Results. The side and the spinal level manipulated did not influence 3D ranges of motion. The mean ranges of motion obtained were 30° axial rotation, 46° lateral bending and 2° flexion. A significant difference of the flexion–extension range existed between manipulations with and without audible release. Axial rotation and lateral bending ranges were correlated. Except for lateral bending which was close to active range, the motion ranges obtained during manipulation were well below active range of motion reported in literature. Conclusions. The results of this study suggest that for the kind of manipulation applied, maximal amplitude between head and trunk does not exceed physiological active range of motion. The amplitude for rotation, which is generally assumed to involve greatest risks for negative side effects, is significantly lower than during active motion. As the study was performed with one practitioner, this result may only be generalized with care. Relevance In spinal manipulative therapy, extreme range of motion as the result of the forces applied is generally believed to represent a major risk for negative side effects, especially with regard to the cervical spine. With a multiple component technique, amplitudes between head and upper trunk were shown not to differ significantly with regard to the side nor to the spinal level. Recorded ranges of motion did not exceed those reported for active motion in literature.