Child head injury is one of the least characterized problems in automotive biomechanics. The cost within the United States exceeds $10 billion per year and it is responsible for 30% of all childhood injury deaths (James, 1999;​ Kraus et al., 1990;​ Ommaya et al., 2002). The common tool for studying this problem is child anthropomorphic test devices (ATDs). However, current child head ATDs are standardized around length, width and height measurements with little attention given to the contour of the skull (Irwin and Mertz, 1997;​ Mertz et al., 1989;​ Weber et al., 1985). The goals of this study were to produce pediatric skull contour data that can be used for the design of child ATDs and finite element models (FEMs) and to investigate methods of providing average skull contours. We obtained CT scans with patient information removed according to HIPAA standards (OCR, 2003) and we developed an orthographic viewer using MATLAB 7.0 (The Mathworks, Inc.) to collect individual skull contours from 3-years-olds and 5-years- olds. The contours were then averaged together using five different averaging techniques. The results showed that the average skull contour data is insensitive to the averaging technique used and that average three-dimensional (3-D) contour data could be produced for child skull ATD design and FEMs. No meaningful statistical differences were found between male and female 3-year-old skull contours, while differences were found at the occiput and in the frontal bone of the contours for the male and female 5- year-olds. The skull contours will be available from the Duke University Injury and Orthopaedics Biomechanics website:​ http:​//biomechanics.bme.duke.edu