Numerous research questions require in vitro testing on lumbar spine and pelvis specimens. The majority of test setups apply forces and torques via the uppermost vertebral body with the lowermost vertebral body fixed and have been validated for kinematics and intradiscal pressure. Models without simulation of muscle traction may produce valid data only for testing conditions for which they have been validated. In vitro test setups with simulation of muscle traction would appear to be useful for conditions beyond such conditions. The aim of the present study was to describe and validate a test rig for the lumbar spine that applies the forces directly to the vertebral bodies via artificial muscle attachments and thus includes the stabilising effects of the muscles known from the literature. The artificial muscle attachments were chosen to get a stable fixation of the pulleys on the cadaver. The location of force application was as close as possible to the physiological footprint of the muscle on the bone. Three paired muscles were combined by individual linear actuators and simulated under force control (posterior, anterior and oblique trunk muscles). An optical 3D motion capture system (GOM, Zeiss, Germany) was used to measure the reposition of the entire lumbar spine and the sacrum against the ilium. At the same time, the force applied to all simulated muscles was recorded. All muscle attachments could be loaded up to a maximum force of 1 kN without failure. The following reposition of the lumbar spine could be generated by the simulated muscle traction keeping the force below each muscle's individual strength: extension 18°, flexion 27°, lateral bending 33°, axial rotation 11°. The effects on lumbar spine reposition of the individual trunk muscles differed depending on the direction of movement. The anterior trunk muscles were the most acting for flexion/extension, at 0.16 ± 0.06°/N, while the oblique trunk muscles were the most acting for lateral bending (0.17 ± 0.16°/N) and axial rotation (0.10 ± 0.14°/N). The maximum nutation of the sacroiliac joint (SIJ) was on average 1,2° ± 0,2°. The artificial muscle attachments to the vertebral bodies proved to be withstand physiologically occurring forces. The range of motion generated in the test rig was physiological. The SIJ nutation determined and the direction of action of the muscle groups correspond to literature data. The order of the individual muscle effects on lumbar spine reposition corresponds to the distance between the muscle insertions and the physiological centre of rotation. In conclusion, taking into account the limitations, the lumbar spine test rig presented here allows the analysis of movements of the lumbar spine and pelvis resulting directly from simulated muscle tractions and thus enables a test environment close to in vivo conditions.