The reliability of repeated measurements of vertical ground reaction force (GRF) variables was determined using 10 clinically sound adult dogs walking on a treadmill three times daily on three different days. Peak vertical force, vertical impulse and duration of stance phase of each limb were determined and analysed for significant differences within and between days and between dogs. Contributions to the variance of GRF were determined using a random effect model including dogs, different days, trials and extremities. ANOVA did not show any significant differences in variables within and between days, but did reveal differences between dogs. Variance was mainly seen in different dogs, whereas different days and trials were less important factors. It was concluded that the ground reaction forces determined by a treadmill system are suitable and provide reproducible parameters in canine gait analysis. Such treadmill/force plate systems allow the simultaneous measurement of all four legs of dogs at a constant velocity walk.