This authors describe a more efficient and adaptive high frequency noise suppression method in which a new adaptive threshold technique is combined with a continuous thresholding function to overcome the shortcoming that existing threshold de-noising technique by wavelet transform is not suitable for seismic data.The continuous hard thresholding function can combine both advantages of soft thresholding function and hard thresholding function,so it can enhance the fidelity of reconstructed signal and reduce the artificial noise.An adaptive threshold scheme is carried out by analyzing the statistical parameters of wavelet subband coefficients like standard deviation,arithmetic mean and geometrical mean in different subbands,which is based on the time-varying and spatial-varying energy distribution feature of nonstationary seismic signal.This threshold can adjust itself automatically with the variation of wavelet coefficient energy in different subbands to meet the requirement of high frequency seismic noise suppression.The actual seismic data processing result indicates that this method can not only raise the signal-to-noise ratio but also protect thoroughly the steep dip angle reflection event and enhance the fidelity of seismic signal after noise elimination.