A background removing algorithm based on wavelet transform was proposed in this paper as background leads to overestimate of peak intensity in various spectroscopy analysis,which needs to be removed before further data processing.The spectrum and its descendants were decomposed by wavelet transform iteratively until the estimated backgrounds converged.A criterion to determine the optimal iteration times was proposed by comparing errors of consecutive estimated backgrounds.By removing the converged estimated backgrounds,the spectrum was ready for peak intensities estimation.This method was evaluated with simulated and experimental Energy Dispersive X ray Fluorescence(EDXRF) spectra.The proposed method was compared with traditional wavelet transform and polynomial fitting method.The analyzed results showed that the proposed method could be used for the accurate elimination of spectral background,which was applicable for other kinds of spectral pre processing.