OBJECTIVE: To determine the natural variability and symmetry of force platform parameters in normal and cerebral palsy gait. DESIGN: Force platform data has been analysed using coefficient of variation, analysis of variances and symmetry index. BACKGROUND: It is important to establish the natural variability of force platform data so that changes in gait due to natural variability and those due to intervention can be determined. METHODS: Force platform data and walking speed were recorded in 15 normal volunteers and 11 children with cerebral palsy on 3 consecutive days. Five recordings were made of each leg on each day. RESULTS: The force platform parameters in both groups showed greater asymmetry in the horizontal plane. In the cerebral palsy group significant asymmetry was found in all 3 directions. The variation of the vertical ground reaction force peaks (Fz1 and Fz3) and of the anterior-posterior negative peak (Fy3) for both groups was low. Fz3 was the most reproducible force parameter. The other force parameters showed high variability and are not considered reliable measures of gait. The temporal parameters and the walking speed were reproducible measures. CONCLUSIONS: Only selected ground reaction force parameters showed acceptable stability and reproducibility. Therefore caution has to be taken in the selection of these parameters, when assessing the effects of any intervention on cerebral palsy gait.