On the theory of wavelet transform,a digital signal f(x) which can be denoted approximately to an approximation signal a_j^jf(x) and a detail signal (or decomposition signal) d_k^jf(x) is analyzed theoretically,and a method for discriminating and removing noises of precursory digital data, and separating high-frequency and low-frequency information from original recording signals is proposed.In the paper,some samples for the method applying to precursory digital data in electromagnetism are given.The study shows that wavelet analysis is a new available method to deal with digital data in electromagnetic precursors.