Background <p>Posterior hemivertebra resection (HVR) with fusion is a widely accepted technique for congenital early-onset scoliosis (CEOS). However, the optimal fusion length remains debated. This study compares the long-term clinical and radiographic outcomes, as well as complication profiles, of two-level fusion (Fused_2) versus multi-level fusion (Fused_GT2) following single-stage posterior HVR in CEOS patients with minimum 10-year follow-up.</p> Methods <p>We retrospectively reviewed 42 CEOS patients who underwent single-stage posterior HVR and pedicle screw fixation before age 10 between August 2003 and March 2014. Patients were stratified into Fused_2 (<i>n</i> = 15) and Fused_GT2 (<i>n</i> = 27) groups based on the number of fused segments. Demographics, surgical parameters (operative time, estimated blood loss), radiographic measures (main and compensatory curves, apical vertebral translation, coronal and sagittal balance, segmental kyphosis, thoracic kyphosis, lumbar lordosis), and complications (crankshaft phenomenon, proximal junctional kyphosis, neurological events, pulmonary complications, adding-on, and reoperation) were assessed preoperatively, immediately postoperatively, and at final follow-up. Comparisons utilized Welch’s t-test and Fisher’s exact test, with <i>p</i> &lt; 0.05 as significant.</p> Results <p>Mean follow-up was 11.02 ± 1.24 years. Fused_2 demonstrated significantly shorter operative time (150.3 ± 38.6 vs. 199.5 ± 52.2&#xa0;min; <i>p</i> = 0.001) and lower blood loss (262.7 ± 193.2 vs. 322.2 ± 188.5 mL; <i>p</i> = 0.342). Preoperative main curve (28.7°±11.5 vs. 42.6°±15.1; <i>p</i> = 0.002) and compensatory caudal curve (9.3°±6.5 vs. 14.3°±8.6; <i>p</i> = 0.040), segmental kyphosis (10.4°±11.0 vs. 22.5°±15.1; <i>p</i> = 0.005), and thoracic kyphosis (15.9°±11.4 vs. 26.9°±16.1; <i>p</i> = 0.014) were significantly smaller in Fused_2, reflecting selection criteria based on deformity severity. Both groups achieved comparable immediate and final correction rates for main and compensatory curves, and similar improvements in apical translation, coronal/sagittal balance, and lumbar lordosis. Crankshaft phenomenon occurred in 35.7% overall (46.7% vs. 29.6%; <i>p</i> = 0.325), PJK in 14.3% (6.7% vs. 18.5%; <i>p</i> = 0.395), and reoperation rate 16.7% (13.3% vs. 18.5%; <i>p</i> &gt; 0.999), with no significant intergroup differences.</p> Conclusions <p>Both two- and multi-level fusion after posterior HVR achieve sustained deformity correction and acceptable complication profiles in CEOS over ≥ 10 years. Fusion length selection should consider HV location, deformity severity, and kyphotic component to balance operative morbidity with long-term stability. Larger studies are warranted to further clarify the relationship between curve magnitude, kyphotic components, and the extent of fusion through regression analysis.</p> Trial registration <p>This study is a retrospective analysis and was not prospectively registered.</p>