Abstract: Objective: This paper gave a new method about evaluating the quality of Congou black tea by near infrared spectroscopy.Methods: There were 240 test samples,180 samples of them used to be a correction stage as the remaining 60 samples a prediction set.Each model is optimized the best waves of the number of segments and best pretreatment method for model inguse to establish the quantitative prediction model by OPUS 7.0software.The smooth points were 17 and dimension was 1.Results: The model predicted a high accuracy which can be used to predict the sensory quality of Congou black tea.The calibration correlation coefficient( Rc) was 96.07% ~98.80%,correcting root mean square error( RMSEC) was 0.148 ~ 0.419; predictive correlation coefficient( Rp) was90.04% ~98.34%,and the RMSEP was 0.105 ~ 0.357. Each model calibration set and prediction set had a higher degree of fit,the out model's prediction accuracy was higher than the other senses several single factor model.Conclusion: The combination of near- infrared spectra of each prediction model to predict the performance had an excellent organoleptic result which established for Congou black tea review quality evaluation.