Objective Assess therapeutic value of specific yoga poses for thoracic and lumbar adolescent idiopathic scoliosis (AIS) taught in office or Internet. Study Design Nonrandomized control trial: Fifty-six adolescents (mean age 14.0 years; mean Risser 3.0) were recruited from our clinic; 41 did the side-plank, the half-moon and elevated side plank poses as appropriate (treatment group) and 15 did not (controls). Thirty curves were treated in office, 30 via Internet. Curve change was evaluated by blinded serial Cobb angles, and analyzed using Mann-Whitney U, paired t-tests and χ2. Results Mean lumbar and thoracolumbar Cobb angle change was −9.2 (95% CI = −11.8, −6.6) in the treatment group and 5.4 (95% CI = 1.7, 9.0) in controls. Both treatment group improvement and deterioration in controls were significant (treatment group: paired t-test t = −7.1, df = 40, p = .000; controls: t = 3.2, df = 12, p = .008). Mean thoracic Cobb angle change was −7.1 (95% CI = −13.1, −1.2) in the treatment group and 9.3 (95% CI = 4.5, 14.6) in controls. Both changes were significant (paired t-test t = −3.3, df = 21, p = .022 for treatment group; t = 4.5, df = 5, p = .006 for controls). Nine Internet patients were non-compliant vs. 6 office patients. Office patients improved 1.6 degrees/month or 5.5%/month; Internet patients improved .72 degrees/month or 3.3%/month. Conclusion These yoga poses show promise for reversing adolescent idiopathic scoliosis. Telemedicine had greater non-compliance and lower efficacy but still produced patient improvement.