The protocol of Davis is widely used in children's gait analysis, especially in cerebral palsy studies and its repeatability was evaluated primarily for adults. The aim of this research was to evaluate the uncertainty and repeatability of this protocol for children. 56 asymptomatic children aged 5-15 years performed the gait exam. Kinematics parameters and Gillette Gait Index (GGI) were calculated. 17 subjects performed the exam twice with markers replacement. Uncertainties on gait parameters were evaluated using repeatability study and Monte Carlo simulations. Uncertainty (2SD of test-retest differences) obtained on angles calculated by the protocol varied between +/-2 degrees and +/-3 degrees (for pelvis and hip in sagittal and frontal planes) and +/-14 degrees (for mean hip rotation). Uncertainty on GGI was +/-12 for healthy subjects. Monte Carlo simulations on 30 cerebral palsy children showed that the error on GGI could reach +/-100 and was correlated to GGI value (R2=0.92): 2SD=24+0.09xGGI.