PURPOSE: We aimed to recognize radiographic and clinical prognostic factors of scoliotic curve behaviour after bracing.METHODS: Our prognostic study was based on the 25 years outcomes of a Boston braced AIS cohort between 1978 and 1993 that were previously reported. All patients were followed-up during bracing, at short term and 25 years post-bracing. We evaluated the impact of socio-demographic, clinical and radiological parameters on the loss of curve correction after bracing.RESULTS: Seventy-seven patients were reevaluated at 25 years post-brace. The mean scoliotic curve was significantly increased after bracing until the 25 years follow-up (p < 0.001). The mean loss of curve correction between the end of bracing and long-term follow-up was independent on the curve type, apical vertebra, premenarcheal status at bracing, time and duration of bracing, Cobb angle before or after bracing. The mean apical vertebral rotation after bracing was significantly related to the loss of curve correction (Spearman ρ = 0.2, p = 0.049). Apical vertebral rotation (Perdriolle method) greater than 20° post-bracing had a three times higher chance of progression > 5° compared with lesser apical vertebral rotation (OR 3.071, CI 0.99-9.51). The rotation of the apical vertebra, type and magnitude of the scoliotic curve after bracing explained 27.4% of the variance in the loss of curve correction post-bracing (R square = 0.274, p < 0.001).CONCLUSION: A scoliotic curve is expected to lose some correction after bracing. The apical vertebral rotation post-bracing mainly affected the long-term curve behaviour. Adolescents with apical vertebral rotation greater than 20° after bracing may need further attention.LEVEL OF EVIDENCE: Prognostic study, Level II.