Abstract: The maximum frequency waveform of Doppler signals from the carotid artery was analyzed using a multi- scale wavelet transform for its advantage in the signal's transient property characterization. A novel index was extracted from the variation of the maximas of the wavelet transform modulus under the different scales. The results of the clinical application showed that this index has a better effect in the cerebral vascular disease diagnosis than the traditional indices from the spectrogram and the indices from the zero-pole model.