Abstract: Abstract: Anthocyanins are a widespread source of natural pigment of foods. Due to their bright attractive color, high solubility in water and health benefits, anthocyanins could be a replacement for synthetic colorant. In recent years, research on the preparation and purification of anthocyanins becomes the focus because of the requirements of bioactive and quantitative analyses. However, the effective methods for the purification of anthocyanins from natural products on a large scale are rarely reported. In this study, isolation and identification of anthocyanins from Vitis davidii Foex skin were investigated. Anthocyanins were extracted from Vitis davidii Foex skin with 70% ethanol containing 0.03% HCl, and then purified orderly by macroporous resin HP-20 column (eluted by 80% ethanol containing 0.05% HCl), polyamide resin column (eluted by 50% methanol containing 0.1% TFA), Sephadex LH-20 column chromatography (eluted by 50% methanol containing 0.1% TFA). The anthocyanins were detected by the ultraviolet-visible (UV-vis) spectrophotometer and the high-performance liquid chromatography (HPLC) after dissolved in methanol containing 0.1% HCl, and then characterized by MSE method using UPLC-QTOF. Moreover, the relationship between the fluorescence spectrum and the structure of the anthocyanins was explored by fluorimetry. The results showed that some anthocyanin (elution time was 11.935 min) was absorbed by polyamide resin, causing the proportion of anthocyanin to change, while Sephadex LH-20 could be used to separate and purify the anthocyanin extracts from Vitis davidii Foex skin. In the process of separation, the column bed must be smooth and tight. The optimum conditions of Sephadex LH-20 column included that the volume of sample was 2 ml and the 50% methanol with 0.1% TFA was used as eluant at a flow rate of 1.5 mL/min. Four types of pigments were isolated from Vitis davidii Foex skin, which were Pigment I, II, III and Ⅳ, and the scanning result of UV-vis showed that Pigment II was not the anthocyanin. Furthermore, Pigment I was identified as malvidin-3, 5-O-diglucoside by the MSE method using UPLC-QTOF, which had a 655 molecular ion peak and 2 detected fragments at 331 and 493. Due to the lack of standards to compare, Pigment III was speculated as malvidin-3, 5-O-diglucoside- -coumary, which had a molecular ion peak at 801 and 3 detected fragments at 331, 493 and 639. Pigment Ⅳ was speculated as the mixture of delphinidin-3-rutinoside and malvidin-3-rutinoside. The purities of Pigment I and Pigment III were determined by analytical HPLC, and estimated to be 98.64% and 98.33%, respectively. Conventional fluorescence detection showed that if the fluorescence was detected under UV light at 254 nm, the pigment in the 5th position would be substituted, but fluorescence spectrum of anthocyanin had not been reported. The relationship between the structure of anthocyanins and the fluorescence spectrum was explored, and it was found that the anthocyanin with diglucoside had 2 obvious fluorescence peaks detected, while the anthocyanin with glucoside had only a fluorescence peak detected. From the result it is inferred that the number of glycosides is associated with the fluorescence intensity and number. The result in this paper provides the reference for detecting the structure of anthocyanins by fluorescence analysis.