By combining wavelet transform with frequency spectrum analysis, wavelet frequency spectrum analysis of an oscillogram was achieved, based on the wavelet property of extracting useful information. Compared with Fourier frequency spectrum analysis of an oscillogram, this method presents many advantages. For instance, explicit peaks in spectrum for the analysis of the sample, easily determined measurable conditions and simple measuring procedure. More important, a novel method, frequency band differential revision method, is described, with which the effects caused by some random factors in the experiment can be substantially eliminated.