Aims: The accuracy of percutaneous CT-Fluoro navigation is compared with the accuracy of the surface-matching procedure. Methods: 68 transpedicular and transvertebral canals were placed percutaneously in an in vitro. The deviation between probe-position and pre-planed trajectory was measured. Evaluated were the mean deviation of the entry point, the exit point, the transverse trajectory angle deviation and the cranio-caudal trajectory angle deviation. Next the soft tissue was removed and the same procedure was done using CT-based surface matching navigation with a registration root mean square of < 1.0 mm. Results: For CT-Fluoro the mean deviation of the entry point was 1.9 mm ± 0.8 (range 0.1–3.2 mm), the mean exit point deviation on the anterior vertebral cortex was 2.1 mm ± 1.1 (range 0.2–3.8 mm). The measurement after surface matching resulted in 1.5 mm ± 0.6 (range 0.0–3.0 mm) for entry point deviation, 1.9 mm ± 0.9 (range 0.1–5.0 mm) for exit point deviation. Conclusions: There is no statistical significant difference of the accuracy between both procedures (Students T-test). Tissue trauma can be reduced as the posterior surface of the vertebra needs not to be exposed as for contemporary registration methods. This offers new promising aspects in percutaneous and minimally invasive spinal techniques