Background and purpose: 3-dimensional midfoot motion is hard to evaluate in clinical practice. We present a new computed tomography (CT)-based radiostereometric analysis (CT-RSA) technique to examine in vivo midfoot kinematics during single-leg stance and compare it with marker-based radiostereometry (RSA).Patients and methods: 8 patients were examined with bilateral non- and full-weight-bearing CT images of the midfoot. 1st tarsometatarsal motion was analyzed using a surface-registration technique (CT-RSA). As all patients had unilateral tantalum markers in the 1st cuneiform (C1) and 1st metatarsal (M1), comparison of precision with markerbased RSA was performed. CT-RSA precision was evaluated with surface registration of both C1–M1 bone and C1–M1 tantalum markers, while RSA precision was determined with C1–M1 markers only. Additionally, to remove motion bias, we evaluated intrasegmental CT-RSA precision by comparing proximal with distal part of M1.Results: Under physical load, the primary movement for the 1st tarsometatarsal joint was M1 dorsiflexion (mean 1.4°), adduction (mean 1.4°), and dorsal translation (mean 1.1 mm). CT-RSA precision, using surface bone or markers, was in the range of 0.3–0.7 mm for translation and 0.6–1.6° for rotation. In comparison, RSA precision was in the range of 0.4–0.9 mm for translation and 1.0–1.7° for rotation. Finally, intrasegmental CT-RSA precision was in the range of 0.1–0.2 mm for translation and 0.4–0.5° for rotation.Conclusion: CT-RSA is a valid and precise, non-invasive method to measure midfoot kinematics when compared with conventional RSA.