Within the area of impact biomechanics, children head models are typically obtained by scaling down adult head models. It is well known however that a child head is not a small adult head. Several anatomists have studied the evolution of skull dimension for male and female since birth. Those studies gather a number of information concerning diameters, volumes, skull thickness (Roche (1953)& Dekaban (1977)). Parallel to those studies Mertz and al (1984) developed a method to obtain the child head geometry from the adult one. Because of the lack of studies concerning children, the "scaling method" seems to be an interesting way to study injury mechanisms for children. A series of scale coefficient are available for different ages, to reduce dimensions, mass, of an adult head to obtain a child head. Data superimposition from both origins allow a comparison and a limitation analysis of the scaling method, at a global (diameters), detailed (shape and thickness) and structural level (presence of trabecular bone, sutures and fontanels). Head geometry of a real child and the one obtained with the scaling method from an adult have also been superimposed for a more indeep comparison. finally, limitations and guidelines for the scaling method are proposed.