Head injury is the leading cause of paediatric death, disability, and Emergency Department visits in the United States. Current paediatric head injury assessment tools are developed primarily based on the scaled adult responses, despite anatomical differences between child and adult heads. The goal of this study was to establish a new paradigm for developing and using subject-specific paediatric head finite element (FE) models for predicting head injuries in young children. To develop this new tool, head CT scans from over 100 0–3 year-old children were collected to develop statistical skull geometry models using principal component analysis and regression analysis. A mesh morphing method was developed to rapidly morph a template head FE model to a target geometry predicted by the geometry model with a given age and head circumference . Age-dependent brain material properties were also quantified using infant and adolescent porcine brain tissues under compression and shear loading conditions. The subject-specific head models were used to reconstruct 50 paediatric cadaver drop tests conducted previously to develop skull fracture risk curves. Moreover, in-depth investigations of 60 paediatric fall cases have also been conducted for computational reconstruction to ensure the accuracy of the model. The results showed that, with this new injury assessment tool, a fall case can be quickly reconstructed in reasonable accuracy, which can provide additional and objective evidence to determine the true causes of a paediatric head injury. The methods developed in this study can enable a rapid development of a complicated FE head model for children accounting for subject-specific head anatomy, and the experiences gained in the in-depth fall investigation and reconstruction significantly enhanced the capability and accuracy of the models. The modelling paradigm will provide a more objective, accurate, and cost-effective way for paediatric head injury prediction in various impact conditions.