Adolescent idiopathic scoliosis (AIS) is a noticeable spinal deformity in both adult and adolescent population. In majority of the cases, the gold standard of treatment is surgical intervention. Technological advancements in medical imaging and 3D printing have revolutionised the surgical planning and intraoperative decision making for surgeons in spinal surgery. However, its applicability for planning complex spinal surgeries is poorly documented with human subjects. The objective of this study is to evaluate the accuracy of 3D printed models for complex spinal deformities based on Cobb angles between 40° to 95°.This is a retrospective cohort study where, five CT scans of the patients with AIS were segmented and 3D printed for evaluating the accuracy. Consideration was given to the Inter-patient and acquisition apparatus variability of the CT-scan dataset to understand the effect on trueness and accuracy of the developed CAD models. The developed anatomical models were re-scanned for analysing quantitative surface deviation to assess the accuracy of 3D printed spinal models. Results show that the average of the root mean square error (RMSE) between the 3DP models and virtual models developed using CT scan of mean surface deviations for the five 3d printed models was found to be 0.5±0.07 mm. Based on the RMSE, it can be concluded that 3D printing based workflow is accurate enough to be used for presurgical planning for complex adolescent spinal deformities. Image acquisition and post processing parameters, type of 3D printing technology plays key role in acquiring required accuracy for surgical applications.