By means of discriminant analysis, it can be identified which class the individuals (data) desired to be classified. In this study, it was aimed to identify that the milk samples belong to which breed, applying discriminant analysis to the milk samples taken from the cattle breeds (19 heads of Jersey, 53 heads of Holstein and 27 heads of Charolaise) in the first lactation. As a conclusion of the study, proper classification actualized in the rate, which can be considered quite high like 90.6%, according to the milk components (fat, protein, lactose, density, pH and conductivity) of Holstein Cattles. This rate actualized as 63.2% in Jersey and 25.9% in Charolaise. It was identified that the rate of general classification of non-linear analysis, used in discriminating Holstein, Jersey and Charolaise breeds from their milk components, was 67.7%.