Abstract: In the fluorescence analysis of pesticides, the fluorescence intensity corresponding to the maximum peak is always chosen as the research subject. In order to obtain the fluorescence intensity of the measured pesticide more accurately and further improve the detection accuracy of the pesticide residues in food, the wavelet transformation is adopted to conduct the denoising process for the measured fluorescence spectrum of the pesticide. In view of the disadvantages of the softthreshold and the hardthreshold methods, a new threshold method, i.e. softandhardthreshold tradeoff method is introduced. The denoised fluorescence spectrum via the wavelet transformation using this threshold method is compared to the denoised spectrum via the analog lowpass filter. The result shows that the denoising spectrum via the wavelet transformation keeps more information for the original signal. Thus the detection accuracy of the system will be improved with the aid of this method.