The support vector machine achieved promising results, outperforming the multiple regression and neural network methods. Such model is useful for understand- ing how physicochemical tests affect the sensory preferences. Moreover, it can support the wine expert evaluations and ultimately improve the production.