The objectives of this study were to characterize organic, biodynamic, and conventional purple grape juices (n = 31) produced in Europe based on instrumental taste profile, antioxidant activity, and some chemical markers and to propose a multivariate statistical model to analyze their quality and try to classify the samples from the 3 different crop systems. Results were subjected to ANOVA, correlation, and regression analysis, principal component analysis (PCA), hierarchical cluster analysis (HCA), soft independent modeling of class analogy (SIMCA), and partial least-squares discriminant analysis (PLSDA). No statistical significant differences (P > 0.05) were observed among juices from the 3 crop systems. Using PCA and HCA, no clear separation among crop systems was observed, corroborating the ANOVA data. However, PCA showed that the producing region highly affects the chemical composition, electronic tongue parameters, and bioactivity of grape juices. In this sense, when organic and biodynamic were grouped as "nonconventional" juices, SIMCA model was able to discriminate 12 out of 13 organic/biodynamic juices and 17 out of 18 conventional juices, presenting an efficiency of 93.5%, while 11 out of 13 non-conventional and 100% conventional grape juices were correctly classified using PLSDA. The use of electronic tongue and the determination of antioxidant properties and major phenolic compounds have shown to be a quick and accurate analytical approach to assess the quality of grape juices.