Abstract: A non-overlapping frequency approach to analyze the time-frequency characteristics of vibration signals is presented by means of the discretized version of continuous wavelet transform based on a modified L-P (littlewood-paley) wavelet basis function, and it is used to study the time-frequency characteristics of blast vibrations in this work. Energy density distributions of vertical ground velocities for 80 kg TNT explosion on the ground surface varying with time show that at the moment of peak particle velocity arriving, the frequency range of blast vibrations is very wide, being the relatively narrow bands at the other moments, and the peak time-frequency energy just corresponds to the peak particle velocity arriving time of blast vibration. The power spectra of blast vibrations through wavelet transform method are agreement with those from Fourier transform. Additionally, the principle of modeling blast vibrations as random evolutionary processes by using the wavelet transform results is outlined.