This study used a previously tested foot model and adapted it for use with children. A number of variations on this adapted model were implemented and tested for repeatability and accuracy on 15 healthy children on three occasions. These included redefinition of the long axes of the tibia and forefoot, assessment of the flexibility of the forefoot and evaluation of the variability of the wand marker on the heel for both static and dynamic trials. It was found that variations on the model produced only minimal changes in repeatability, the only significant change being elimination of the wand marker on the heel in the static trial, which reduced between-day variability of hindfoot motion in the transverse plane. However, some differences were evident in the mean values for all variations. Based on these results, the most accurate and appropriate version of the model is proposed, and average kinematic curves are presented based on the measurements from 14 healthy children.