Abstract: Abstract: Infrared spectrum (IR) reflects the molecular structure of unknown material, and responds with varied chemical bonds. So it is used to determine chemical composition contents and evaluate product quality in livestock products extensively. Milk is a key part in human nutrition intake. And the exact determination of nutrients and the proper evaluation of quality have important significance for milk production. This paper introduced the application of IR in each link of milk production. Fat and protein contents in milk vary in different dairy farms, and many factors affect milk quality, which contribute to the final price of raw milk in acquisition. Milk composition determination using IR is likely to give a quick and comprehensive evaluation for milk quality. Unknown and undeclared adulterants of milk threaten consumers' health seriously. Qualitative and quantitative analysis models provide a convenient identification method for milk adulteration based on spectrum variation of adulterants. Milk trait related to cow health and robustness is very important for dairy farm management. Diagnosis of ketoacidosis and body energy status using IR instruments is helpful for accurate breeding in dairy farm. This paper reviewed recent literatures in order to evaluate the general trends of infrared spectroscopy application on milk production. On the basis of introducing the data processing and model building, this paper presented a review of the overseas and domestic researches on milk composition and milk coagulation properties using IR, especially for milk protein fraction and fatty acids composition. We compared the model performance of optical spectroscopy from different research reports. The effects of reference method, sample size and unit on model parameters were discussed in particular. Moreover, IR was efficient for phenotypes assessment and genetic selection based on these models. The variances of absorption on IR caused by adulterants spiked in milk not only indicated the appearance of milk adulteration, but also displayed the difference between cow milk and soy milk. Milk spectrum was proved to be heritable in specified wavelength, while some other bands varied with different enviromental factors. And many literatures confirmed the correlation between cow's feed and milk optical characteristic. Although nonnegligible random error and data variability existed in sampling, IR reflected energy status of dairy cows with moderate accuracy. Mid-IR has been also studied as a potential tool to predict several milk traits related to cow health, such as ketone bodies, which were closely related to cow fertility and production. IR was also used to predict methane emissions from cow digestive tract. The advantages of infrared spectroscopy analysis were emphasized, and we also listed potential challenges existing in instrument setting, data collection and model building. The objective of this paper was to highlight the application of infrared spectroscopy on milk traits, which was related to milk composition and quality, and dairy farm management. Considering the overall trends, we proposed some future research directions of this methodology on milk production, including prediction of trace nutrients, uniformity of references methods and units, possibility of spectrum assessment, and diagnosis of disorder and fertility. With the future developments in these areas, infrared methods would be more popular in milk composition determination, quality control, and dairy farm managements, with higher accuracy, efficiency and convenience.