In order to achieve non-destructive of mature vinegar varieties, a fast discrimination method was put forward based on Visible_near infrared reflectance (NIR) spectroscopy. The FieldSpec3 spectrometer was used for collecting 20 sample spectra data of the three kinds of mature vinegar separately. Then principal component analysis (PCA) was used to process the spectral data after pretreatment using average smoothing method and multiplicative scatter correction (MSC) method, and principal components(PCs) were selected based on accumulative reliabilities. A total of 60 mature vinegar samples were divided into calibration sets and validation sets randomly, the calibration sets had 45 samples and the validation sets had 15 samples. The stepwise discriminant analysis was trained with five PCs in calibration sets as the inputs,and mature vinegar varieties as the outputs. The stepwise discriminant analysis model was built for discrimination of mature vinegar variety ,and the model contains 15 samples in the validation sets. The result showed that a 100% recognition ration was achieved.The BP-ANN model for discrimination of mature vinegar varieties were built based on PCA and the stepwise discriminant analysis, then the model was tested with the 15 sample in the validation sets. The result showed that a 100% recognition ration was achieved with the threshold predictive error ±0.027. It based on five principal components had a higher prediction accuracy and efficiency more than the BP neural network model. It could be concluded that PCA combined with stepwise discriminant analysis and BP-ANN was an available method for varieties recognition of mature vinegar based on NIR spectroscopy.