OBJECTIVE : Bracing is the most commonly used treatment for scoliosis. But braces remain predominantly "handcrafted." Our objective was to create a novel brace simulator using a high-fidelity 3D "avatar" of the patient's trunk. METHODS : An observational cross-sectional study was constructed. The inclusion criteria were patients with a moderate idiopathic scoliosis (between 15° and 35° of Cobb angle) aged between 9 and 15 years old with an indication of brace treatment. Twenty-nine scoliotic patients, 25 girls and four boys, with a mean age of 12.4 years were included. Twenty right thoracic and 14 left lumbar were measured with a mean Cobb angle of 24°. 3D "avatars" were generated using a novel technology called the "anatomy transfer." Biomedical simulations were conducted by engineers who were blinded to the clinical effect of the real patient brace. The in-brace Cobb angle effect (real effect) was compared with the virtual numeric in-brace Cobb angle observed using the blindly constructed avatar (simulation effect). RESULTS : Real and simulated in-brace Cobb angle were compared using a paired two-sided Student's t test. The real mean Cobb angle was 11° and 17° in the simulation which was statistically significant. The strength of prediction of the simulation was assessed for each individual patient; 76% of the real in-brace Cobb angles had good and moderate prediction (± 10°). CONCLUSIONS : Incorporating high-fidelity copy of the entire 3D shape of the patient's trunk and multiple 3D-reconstructed bony images into an anatomical reference avatar resulted in moderate-to-good prediction of brace effect in three quarters of patients. These slides can be retrieved under Electronic Supplementary Material.