Abstract: The carboxymethyl cellulose( CMC) adulterated in konjac glucomannan( KGM) was detected by Fourier transform near infrared spectroscopy(FT-NIR). Principal component analysis and partial least squares techniques were used for qualitatively analyzing of the spectra and establishing calibration models from the spectra pretreated by various correction algorithms including baseline correction,smoothing,vector normalization,first derivative and second derivative. The predication accuracy and reliability of the models were checked by cross-validation and external validation. Coefficients of determination(R_c² in calibration set) and root mean square error of the calibration models ranged from 0.933 to 0.997 and 7.64% to 1.56%,respectively,depending on the mathematical pre-treatment methods on the spectra. The optimal calibration model with first derivative(5 points) pretreatment was obtained by using the validation set that gave a root mean square error of prediction of 8.37%. The analytical results demonstrated that the correlation between the chemical value(true value) of samples of calibration set and the FT-NIR predicated value was 0.9905. The results indicated that chemometric- assisted FT- NIR spectroscopy could be a simple and efficient tool for the detection and quantification of KGM adulterated with cheaper CMC.