In order to dynamically detect vehicle motion stability during operation, as a new method of detecting vehicle motion stability, the time-varying damping ratio, is presented on the basis of the analysis of the dynamic response feature and the non-stationary characteristics of the response signals. The lateral vibration accelerations are decomposed by the IEMD (improved empirical mode decomposition) into a finite and often small number of IMF (intrinsic mode functions). The instantaneous frequency and the instantaneous amplitude of the intrinsic mode components are drawn through normalized Hilbert transform, respectively. The time-varying damping ratios under different intrinsic modes are computed by applying instantaneous amplitude and instantaneous frequency to judge the running stability of the vehicle. The zero or negative damping ratio indicates that the vehicle is unstable, the positive damping ratio shows that the vehicle keeps stable, the absolute value of the negative ratio means the severity of vehicle instability. Through the theoretical and experimental verification applying simulation and the measured signals, the results show the method can accurately and efficiently dynamically capture the time-varying characteristics of the system stability, with a powerful method for transit monitoring of vehicle running stability. ©, 2015, Editorial Office of Chinese Journal of Mechanical Engineering. All right reserved.