The traditional methods of analyzing main constituents in milk, for example the Kjeldahl method for protein, Röse-Gottlieb or Mojonnier methods for fat, and polarimetry method for lactose, are time consuming and expensive. The near infraredspectroscopy (NIR) method provides a simultaneous quantitation of a number of milk constituents like proteins, fats and carbohydrates. It is a non-destructive and rapid measurement. Sample dilution is not necessary and even comparatively thick samples can be measured. The NIR spectrum consists of combination bands and overtone vibrations of the fundamental middle infrared bands, and mainly –CH, –OH, and –NH bonding vibrations are observed in the NIR region. However, unlike fundamental middle-IR bands, NIR bands are generally weak intensity, broad and overlapping. Because of the high similarity in NIR spectra, chemometrics data analysis such as principle component regression (PCR), multi-linear
regression (MLR) or partial least squares (PLS) regression is required to correlate spectral data with the reference values of measuring components. Here, we introduce a method for quantitative determination of protein, total fat and carbohydrate contents in milk using NIR spectrometry and PLS quantitative calibration method.