Center of mass displacement during gait has frequently been used as an indicator of gait efficiency or as a complement to standard gait analysis. With technological advances, measuring the center of mass as the centroid of a multi-segment system is practical and feasible, but must first be compared to the well-established Newtonian computation of double-integrating the ground reaction force. This study aims to verify that the kinematic centroid obtained from a commonly-used model (Vicon Peak (R) Plug-In-Gait) provides at least as reliable measurements of center of mass displacement as those obtained from the ground reaction forces. Gait data was collected for able-bodied children and children with myelomeningocele who use larger lateral center of mass excursions during gait. Reasonable agreement between methods was found in fore-aft and vertical directions, where the methods' excursions differed by an average of less than 10 mm in either direction, and the average RMS differences between methods' computed curves were 6 and 13 mm. Particularly good agreement was observed in the lateral direction, where the calculated excursions differed by an average of less than 2 mm and the RMS difference was 5 mm. Error analyses in computing the center of mass displacement from ground reaction forces were performed. A 5% deviation in mass estimation increased the computed vertical excursion twofold, and a 5% deviation in the integration constant of initial velocity increased the computed fore-aft excursions by 10%. The suitability of calculating center of mass displacement using ground reaction forces in a patient population is questioned. The kinematic centroid is susceptible to errors in segment parameters and marker placement, but results in plausible results that are at least within the range of doubt of the better-established ground reaction force integration, and are more useful when interpreting 3-D gait data.