Melamine is an illegal food additive in beans and dairy products. As a cheap substitute for protein, melamine was illegally added into milk powder and other foods, causing serious social harm and greatly threatening the safety of people’s lives and property. At present, spectral technology has become an effective means to identify and quantitatively detect illegal food additives, which provides a reliable research method and identification basis for the quality supervision department. The timeliness, non-destructive and accuracy of spectral detection technology improve the detection efficiency of melamine in food and promote the development of accurate and automated food quality detection. In recent years, a large number of studies have focused on the spectral detection of melamine, such as the development of new enhanced substrates or sensors to reduce the detection limit of melamine and improve the detection accuracy; Develop more portable automatic spectrum rapid detection equipment, reduce the detection cost and improve the detection efficiency. These spectral technologies have their advantages, but it is difficult to form a standardized and unified detection specification, making all kinds of spectral detection technologies only stay in the experimental stage and can not be applied to actual combat. On the other hand, with artificial intelligence and pattern recognition technology, spectral data analysis methods have made great progress in recent years. Various spectral preprocessing and data modeling methods have been proposed, which greatly improves the sensitivity and stability of spectral detection technology. The application status of spectral techniques (Raman spectroscopy, near-infrared spectroscopy, fluorescence spectroscopy, spectral imaging, etc.) in the detection of melamine in recent ten years was reviewed. The detection limits, quantitative ranges and sample pretreatment methods of different instruments were summarized; The applicability of various spectral preprocessing and spectral data modeling methods in different spectral data is analyzed, the advantages and disadvantages of these methods and the suitable instruments are summarized, and their application prospects and research trends have prospected.