This paper proposes a method of signal analysis for a semi-active suspension isolation system. A semi-active suspension isolation system is an efficient equipment for vibration isolation and noise reduction in a car. In order to achieve a good vibration isolation effect, it adjusts its damping characteristic according to the excitation signals caused by different road conditions. Thus the key technique of the control of a semi-active suspension isolation system is to classify the excitation signals and analyze them quickly in time. This paper puts forward a signal processing method for these signals. At first, these signals are subdivided into three types, the stationary signal, the non-stationary signal and random signals. Subsequently, it uses the short-time Fourier transform (STFT) to handle them with different window functions in order to find the main frequency of the signal. Especially, the dynamic trend of the main frequency has to be considered for non-stationary signals. The two main parameters of the main frequency of the excitation signal, namely the position and the amplitude, will be obtained by the signal analysis. Then the control system can use them to give the best damping ratio for the damper. Finally, the method is used in a two-stage semi-active vibration isolation system to verify its effect. The experimental results show that it can analyze the signal exactly in time, and obviously enhance the vibration isolation performance of the semi-active suspension isolation system.