In order to distinguish figs of different maturity accurately and quickly without damage, the samples of figs were collected by near-infrared spectroscopy, and K-means clustering was carried out for five indexes of fig sugar degree, single fruit weight, vertical diameter, horizontal diameter and hardness. According to the spectral data and the score chart of principal component analysis, the components of the optimal clustering effect and the index distribution of each category were determined. Based on the PLS-DA model, the cluster discrimination model was constructed to achieve the purpose of fruit maturity classification. Through the analysis of the differences of the above five indexes, significant differences in soluble solids, single fruit weight and hardness of the three kinds of fig samples at the maturity stage were found, and significant differences between the vertical and horizontal diameters of mature and growing fruits and young fruits were also detected. According to PLS-DA discriminant model, the classification accuracy of training set was 99.59%, and that of test set was 99.15%. The results showed that the PLS-DA model based on principal component analysis and spectral data had good performance, and it could be used to identify the maturity of figs quickly.