Purpose <p>This study aimed to identify and validate prognostic factors to predict scoliosis progression.</p> Methods <p>One hundred sixty-two girls, aged 13.5 ± 1.7 years old, diagnosed with idiopathic scoliosis were recruited. One hundred were used for model development, and 62 for testing, in which the number of progression cases was 25 and 11, respectively. All participants were scanned by an ultrasound (US) system for two consecutive visits (baseline and follow-up). The baseline parameters included (a) demographic: age, body mass index (BMI), and menarche status; (b) radiographic: X-ray Cobb angle, number-of-curve (NOC), and Risser sign; (c) ultrasonic: US Cobb, maximum axial vertebral rotation (AVR), Cobb angle at the plane of maximum curvature (PMC), kyphotic angle (KA) and reflection coefficient (RC) index. For the follow-up visit, only the US Cobb was recorded. The demographic and X-ray parameters were extracted from the scoliosis clinical records. The US parameters were measured by trained raters with good measurement reliability. The prediction model was developed using logistic regression analysis.</p> Results <p>The final predictors were US Cobb change, RC index, and NOC, and the probability of curve progression risk was Log (p/1-p) = -1.40 + 0.28(US Cobb change)-39.45(RC) + 1.36 (NOC). The model achieved sensitivity, specificity, and accuracy of 90% on the 62 test dataset, which was much better than using US Cobb change only.</p> Conclusions <p>The results indicated that participants with lower RC (weaker bone, RC ≤ 0.06), larger US Cobb change (≥ 5°), and multiple curvatures (NOC &gt; 1) were at a higher risk of curve progression. A large study including more progression cases is needed for further validation.</p>