Purpose <p>Compared to linear models, segmented models allow for the identification of threshold effects in severe spinal deformities, beyond which patients’ health-related quality of life (QoL) declines significantly. However, no study has established threshold values for cosmetic parameters in adolescent idiopathic scoliosis (AIS). This study aims to determine the threshold values of cosmetic parameters in spinal deformity severity measurements.</p> Methods <p>A total of 278 female adolescents with idiopathic scoliosis (age: 14.7 ± 2.0; maximum Cobb angle: 49.4° ± 15.9°) completed the SRS-22r questionnaire. The Cobb angle and cosmetic parameters were measured using X-rays and Image-Pro Plus 6.0 software. Segmented and linear regression models were used to evaluate the correlation and explanatory power (<i>R²</i>) between SRS-22r domains and spinal deformity indicators, and to identify the thresholds of cosmetic parameters that significantly impact SRS-22r scores.</p> Results <p>Lateral shoulder tilt (LST) predicted significantly more variance in all SRS-22r domains using segmented (<i>R</i>^<i>2</i>: 0.01–0.08) than linear models (<i>R</i>^<i>2</i>: 0.00–0.03). Segmented models with a threshold estimated at 5.2° of LST explained 1–6% more variance than the corresponding linear models using the same variables. Cobb angle did not strongly associate with SRS-22r total score variables with linear and segmented models, explaining less than 4% of the variance.</p> Conclusion <p>This study shows that QoL stays relatively stable until LST exceeds a threshold of 5.2°, after which it drops markedly. Although the link between QoL and cosmetic parameters is weak, segmented models provide a better explanation than linear models.</p>