Objective:To explore the effect of conducting core stability training (CST) on an unstable supporting surface using thoracolumbar fracture patients with an incomplete spinal cord injury.Methods:Forty thoracolumbar fracture patients with incomplete spinal cord injury were randomly divided into an experiment group and a control group, each of twenty. Both groups received 30 minutes of CST five times per week for 8 weeks. The patients in the control group were trained on a stable supporting surface while those in the experiment group used an unstable surface. Evaluations were conducted before and after the 8-week intervention. Gait and static balance data were collected and analyzed using 3D motion analysis software and an EAB-100 active balancer.Results:After the intervention, the average stride length and comfortable walking speed of the experimental group were both significantly better than the control group′s averages. Moreover, the path length, circumferential area, rectangular area and effective value area of the Romberg rate were all significantly better, on average, in the experiment group, as was the average displacement of the deflection center with the eyes closed in static balance.Conclusions:An unstable supporting surface is superior to a stable one for conducting CST after thoracolumbar fracture with incomplete spinal cord injury. The effect may be due to improved nonvisual postural control.