Objective To investigate the gait stability of healthy old and young adult volunteers during walking using the nonlinear time series analysis method so as to comprehensively assess the dynamic balance of human and provide important references for the prediction of fall risk. Methods The Vicon motion capture system was used to collect three-dimensional kinematic data of healthy volunteers (seven old subjects and ten young subjects) at different walking speed (80%, 100%, 120% of the natural walking speed). The anterior-posterior and medio-lateral motions of the seventh cervical and tenth thoracic vertebrae, as well as the flexion-extension and abduction-adduction angles of the lower extremity joints, were obtained from 30 consecutive gait cycles to calculate the largest Lyapunov exponents and analyze the difference of gait stability between the old and young group and the influence of walking speed on gait stability. Results The gait stability in the young group was better than that in the old group, and there were significant differences in certain motion segments (P＜0.01). The gait stability in both groups were reduced with the increase of speed (P＜0.05). Conclusions The largest Lyapunov exponent based on nonlinear time series analysis method can be used to effectively and quantitatively analyze the gait stability of each motion segment in human during walking.