Fusion level selection in Lenke 5 adolescent idiopathic scoliosis (AIS) remains debatable, with no consensus regarding selective, hyperselective or non-selective thoracolumbar/lumbar fusion. Supine side-bending (SSB) radiographs are used to assess main thoracic (MT) curve structurality and differentiate Lenke 5 from Lenke 6 curves. However, this assumes that end vertebrae identified on standing radiographs remain representative during side bending. Emerging evidence suggests some Lenke 5 curves exhibit thoracic behaviour resembling Lenke 6, indicating occult MT structurality not captured by standard SSB assessment. Improved characterization of MT behaviour on SSB radiographs may therefore aid fusion strategy selection. This study aimed to evaluate Main Thoracic Internal Cobb (MTICobb), a novel radiographic parameter to assess occult MT structurality and its ability to differentiate selective and non-selective thoracolumbar/lumbar fusion strategies in Lenke 5 AIS. A retrospective study included 202 Lenke 5 or 6 AIS patients who underwent posterior spinal fusion between 2019 and 2024. MTICobb was measured on right SSB radiographs using the most tilted vertebrae within the MT curve. MTICobb was defined as “present” when the most tilted vertebrae forming the Cobb angle lay internal to the end vertebrae identified on standing radiographs. Correlations between MTICobb and fusion extent as well as UIV were assessed. Receiver operating characteristics (ROC) analysis evaluated MTICobb discrimination of selective versus non-selective fusion. Optimal threshold was determined using the Youden index. MTICobb was present in 88.6% of patients, with similar prevalence in Lenke 5 and 6 groups. In Lenke 5, higher MTICobb correlated with greater fusion extent (r = 0.563, p < 0.001) and more cephalad UIV selection (r = −0.584, p<0.001). MTICobb demonstrated acceptable discrimination between selective and non-selective fusion (AUC = 0.873). A threshold of 21° yielded sensitivity 0.661 and specificity 0.914 to identify patients undergoing non-selective fusion in Lenke 5. In conclusion, higher MTICobb was associated with greater fusion extent and more cephalad UIV selection in Lenke 5. MTICobb demonstrated acceptable discrimination between selective and non-selective fusion, supporting its use as an adjunct in fusion decision-making.