OBJECTIVE: This retrospective study was conducted to analyze the novice neurosurgeon's experience of cervical pedicle screw placement by using the technique with direct exposure of pedicle via para-articular mini-laminotomy. METHODS: Fifteen sawbone models of subaxial spine were used (124 pedicles) to evaluate efficacy of repetitive training improving accuracy of cervical pedicle screw insertion. After that, we retrospectively reviewed 9 consecutive patients presented with traumatic cervical lesion. A total 38 cervical pedicle screws had been inserted. We analyzed the direction and grade of pedicle perforation on the postoperative computed tomography scan, and learning curve by using sawbone model. RESULTS: In sawbone model group, the correct position was found in 102 (82.3%) screws, and the incorrect position in 22 (17.7%) screws. The incidence of incorrect screw position was 26.9% in the initial 9 sawbone model, and 0% after that. Among the 38 screws inserted in 9 patients, the correct position was found in 36 (94.7%) screws, and the incorrect position in a 2 (5.3%) screw. There was no neurovascular complications related with cervical pedicle screw insertion. CONCLUSION: In vitro training to insert pedicle screw by using sawbone models could improve an accuracy of cervical pedicle screw placement by using this technique. Preliminary result revealed that cervical pedicle screw placement would be feasible and provide good clinical results in traumatic cervical lesions.