Background: Previous thorax models have been proposed for gait analysis, however these require markers to be placed on the back. This presents a limitation in the kinematic analysis of the thorax under load carriage conditions. Research question: This study evaluated the validity and reliability of a thorax marker set that does not require markers to be placed on the back (HubemaLab model) when compared to 3 previously published marker set models. Methods: 17 young adults were evaluated while walking at their self-selected speed. A 12 camera motion capture system was used to acquire the marker position data which was then processed using the respective models using Visual-3D. The level of agreement for the flexion/extension peak, right/left lateral peak and right/left rotation peak of the thorax angle and angular velocity; together with the range of motion and thorax angular velocities in the three planes was found between each thorax marker set, while the reliability was measured using the intraclass correlation coefficient. Results: The ICC results for the thorax angle ROM and the range of thorax angular velocity between the HubemaLab model and the other models showed excellent to good reliability in all three planes. While the ICCs for the peak flexion/extension, peak right/left lateral flexion and peak right/left rotation showed excellent to moderate reliability in all three planes. Conclusion: The new model could be potentially valuable for kinematic gait analysis under load carriage conditions which obscure markers placed on the back.