Abstract: In order to deeply understand the impact of clinical mastitis on the physicochemical properties, flavor and taste of cow milk, a milk composition analyzer combined with an electronic nose and an electronic tongue was used to distinguish milk from cows with and without clinical mastitis. Moreover, nuclear magnetic resonance (NMR) spectroscopy was used to identify and quantify small molecules in milk samples. The fat content of mastitic milk was significantly lower than that of milk form healthy cows, while the density, protein and freezing point were significantly higher than those of milk from healthy cows. The electronic nose and the electronic tongue were able to distinguish the differences in the flavor and taste of milk from normal and mastitic milk. A total of 54 small molecules were identified in the two milk samples, including amino acids (17), organic acids (15), carbohydrates (7), nucleosides, nucleotides and their derivatives (5). The concentrations of 23 compounds significantly differed between both milks. A total of 16 predominant compounds were found in normal milk, and seven in mastitic milk. The contents of amino acids and organic acids in mastitic milk significantly increased, resulting in unpleasant sour odor of milk. This study provides a scientific rationale for the discrimination between normal and clinical mastitic milk and for the quality control of raw milk

Key words: milk; flavor substances; taste substances; metabolomics; intelligent sensory detection; mastitis