Objective?To analyze the correlation between adenoid hypertrophy (AH) severity and craniofacial morphological characteristics or clinical symptoms in children, and develop a composite prediction model for severe AH based on clinical symptoms and craniofacial features. Methods?This cross-sectional study enrolled 201 children aged from 6 to 8 years from the Department of Otorhinolaryngology-Head and Neck Surgery, Shanghai Ninth People’s Hospital (August 2023 to December 2024). Lateral craniofacial photographs, clinical characteristics (respiratory patterns, tonsillar hypertrophy, rhinitis, asthma, snoring), and OSA-18 questionnaire data were collected. T The study participants were randomly divided into two groups (70% vs. 30%) for further analysis., including 149 cases in the modeling group and 52 in the external validation group. The modeling group was classified into non-severe AH (n=77) and severe AH (n=72) subgroups via nasal endoscopy. Craniofacial angles/ratios were measured using specialized software. Logistic regression with Lasso regularization identified independent predictors, followed by nomogram model construction. Model performance was evaluated via receiver operating characteristic (ROC) curves, calibration plots, and decision curve analysis (DCA), with external validation assessing sensitivity and specificity. Results Compared to the mild/no AH group, patients with severe AH exhibited significant craniofacial morphological differences: reduced facial convexity (173.24 ± 2.71° vs. 171.01 ± 4.08°; P < 0.001), increased maxillary convexity (30.37 ± 6.52° vs. 35.98 ± 7.25°; P < 0.001), and enhanced lip prominence (0.90 ± 0.09 vs. 1.02 ± 0.11; P < 0.001). Multivariate analysis revealed independent risk factors for severe AH: breathing patterns (OR<1, P<0.05, protective factor), tonsillitis (OR=6.035, P=0.007), severe rhinitis (OR=5.183, P=0.013), asthma (OR=14.927, P=0.002), snoring (OR=5.803, P=0.011), The composite prediction model demonstrated excellent discrimination, with AUC of 0.949 in the modeling cohort AUC of 0.961 in the validation cohort. Calibration curves indicated high consistency between predicted and observed probabilities, while DCA confirmed clinical utility. Conclusion:?Mouth breathing, tonsillitis, rhinitis, asthma, snoring, increased facial/lip convexity, and reduced maxillary convexity are significant predictors of severe AH. The established composite model integrating craniofacial morphology and clinical features exhibits robust predictive performance, offering a valuable tool for early screening and risk stratification of severe AH in children.