Abstract: In order to select a suitable method for preprocessing,the effects of standard normal variate(SNV),multiplicative scatter correction(MSC),derivation calculation(DC) and continuous wavelet transform(CWT) on the prediction results of NIR spectra models for the contents of total sugar,total nitrogen and nicotine in tobacco were investigated.It was found that:1) SNV and MSC could correct the background at different intensities,while DC and CWT could remove the background from NIR spectra;and 2) the prediction results could be improved to a certain extent by either background correction or background removal,and would be better by combining the two.