The glucose, fructose, ethanol and glycerol were predicted by using NIR spectroscopy and multivariate calibration methods. And the reference values were analyzed by HPLC to evaluate four components above. The principle component analysis regression (PCR) and partial least squares regression (PLSR) were compared. The correlation coefficient of calibration (R2), the root mean square error of calibration (RMSEC) and the root mean square error of prediction (RMSEP) were used to evaluate the models. Results from the application of PCR and PLSR were presented, showing there were no significant differences between PCR and PLSR to predict glucose, fructose and ethanol. While for the glycerol, it was better to use the PLSR. PLSR almost always required fewer latent variables than PCR, but this did not appear to influence predictive ability.