Background and objectives Mixed reality (MR) is one of the image processing technologies that allows the user to manipulate three-dimensional (3D) virtual images (hologram). The aim of this study was to evaluate the accuracy of MR-based pedicle screw (PS) placement using 3D spine models. Materials and methods Using the preoperative CT data of a patient with adolescent idiopathic scoliosis (AIS) who had undergone posterior spinal fusion in our hospital, a 3D-printed spine model was created. On the other hand, a 3D hologram of the same patient was automatically created using the preoperative CT data uploaded to the Holoeyes MD service website (Holoeyes Inc., Tokyo, Japan). Using a Magic Leap One® headset (Magic Leap Inc., Plantation, FL), the 3D hologram with lines of predetermined PS trajectories was superimposed onto the 3D-printed spine model and PS were inserted bilaterally along with the trajectory lines from T5 to L3. As a control, we used a readymade 3D spine model of AIS and inserted PS bilaterally with a freehand technique from T4 to L3. The rate of pedicle violation was compared between the MR-based and freehand techniques. Results A total of 22 and 24 PS were placed into the 3D-printed spine model of our patient and the readymade 3D spine model, respectively. The rate of pedicle violation was 4.5% (1/22 screws) in the MR-based technique and 29.2% (7/24 screws) in the freehand technique (P = 0.049). Conclusions We demonstrated a significantly lower rate of PS misplacement in the MR-based technique than in the freehand technique. Therefore, an MR-assisted system is a promising tool for PS placement in terms of feasibility, safety, and accuracy, warranting further studies including cadaveric and clinical studies.