The accuracy and safety of pedicle screw insertion was markedly improved with the introduction of intraoperative three-dimensional navigation system during the last decade. This study aimed to evaluate the accuracy of pedicle screw placement using O-arm-based navigation system versus conventional freehand technique. Methods: We reviewed the accuracy of 341 thoracic (n = 173) and lumbosacral (n = 168) pedicle screws placed in 60 consecutive patients using either O-arm-based navigation or freehand technique in the Department of Neurosurgery of Beijing Tsinghua Changgung Hospital between January 2015 and June 2018. Patient-specific characteristics, treatment-related characteristics, and screw-specific accuracy were analyzed. The accuracy of pedicle screw placement was measured by Gertzbein-Robbins scale and screw grades A and B were clinically acceptable. Results: One hundred ninety-one screws were inserted in the O-arm-based navigation group and 150 in the freehand group. One hundred eighty-three (95.81％) clinically acceptable screws were placed in the navigation group and 135 (90.00％) in the freehand group (p = 0.034). Twenty-three (6.74％) screw revisions were performed in the two groups (8 screws in the navigation group and 15 screws in the freehand group) and significant difference was observed in thoracic spine (p = 0.018), while no statistical significance was presented in lumbosacral spine (p >0.05). Twenty-four (12.57％) screws in the navigation group and 24 (16.00％) in the freehand group violated the cortex (p > 0.05). Medial screw deviation was the most common problem in the two groups. Conclusion: The O-arm-based navigation exhibits higher accuracy for pedicle screw insertion than the freehand insertion technique.