In order to classify the orange juice beverages effectively, the fluorescence character differences of two kinds of orange juice beverages including 100% orange juice and orange drink were analyzed and compared, principal component analysis combined with Euclidean distance was adopted to classify two kinds of orange juice beverages, and ideal classification results were obtained. Meanwhile, the orange juice content estimation model was established by using fluorescence spectroscopy combined with partial least squares regression method, and the correlation coefficient R, root mean square error of calibration RMSEC and root mean square error of prediction RMSEP were 0.997, 0.87% and 2.05%, respectively. The experimental results indicate that the calibration model offers comparatively accurate content estimation, which reflect the actual orange juice content in the commercial orange juice beverages. The exploration to classify orange juice beverages was carried out from two aspects of qualitative and quantitative analysis by employing fluorescence spectroscopy combined with chemometrics method, which can provide a new idea for the classification and adulteration detection of commercial orange juice beverages, and also can give certain reference basis for the quality control of orange juice raw material.