[en] The interest for the dairy products with higher nutritional quality increases. The aim of this research was to elaborate different calibration equations to predict by Mid-Infrared Spectrometry the fatty acid contents in bovine milk. 1,609 milk samples were collected between March 2005 and May 2006 for 475 cows from 6 dairy breeds (Dual Purpose Belgian Blue, Holstein, Jersey, Montbeliarde, Normande and Red and White) in 8 herds. 78 samples were chosen using Principal Components approach based on spectral variability. All samples were scanned by MilkoScan FT6000. The reference fatty acid concentrations were measured by gas chromatography with a capillary column of 100 m length. The calibration with Partial Least Squares (PLS) on 78 samples showed a ratio of standard error of cross-validation to standard deviation (RPD) ranged between 1.5 and 6.76. The FA present in high concentration in milk were better predicted as in previous studies. In conclusion, the development of this fast method to predict the FA contents and directly integrating in the milk recording structure gives new perspectives for the dairy industry to detect easily and nally improve nutritional quality of their dairy products.
