Abstract: In order to prove the effectiveness of 3-D fluorescence spectroscopy in the rapid classification of milk beverages on the market, the 3-D fluorescence spectra of 12 different brands of milk beverages which were brought from the market were collected and the spectral data were pretreated. The spectral data after pretreatment were classified by principal component analysis (PCA), factor analysis, and cluster analysis. Through factor analysis, 8 factors representing 3-D fluorescence spectrum data were extracted, and the data after dimension reduction were analyzed by clustering. The results showed that the cumulative contribution rate of the first two principal components in PCA was 95.15%, which could reduce the dimension of spectral data well, but the classification threshold was not easy to be determined. And by selecting 2.5 as a threshold distance, these milk beverages could be divided into formulated milk beverages, fermented milk beverages and lactobacillus beverages and the classified accuracy can reach above 83%. This study verified the feasibility of 3-D fluorescence spectrum binding factor analysis for classification of milk beverages.