Three different electrodes were tested for use as nonspecific amperometric sensors for blind analysis on real matrices, namely different fruit juices from different fruits or different brands. The first two electrodes were traditional Pt and Au electrodes, while the third one was modified with poly(3,4-ethylenedioxythiophene) conducting polymer. The sensors were tested separately, tested coupled to each other, and also tested together. The responses of the electrode system(s) were first screened via PCA and then their discriminant capabilities were quantified in terms of the sensitivities and specificities of their corresponding PLS-DA multivariate classification models. Particular attention was paid to analyzing the evolution of the response over subsequent potential sweeps. The modified electrode demonstrated the most discriminating ability, and it was the only system capable of satisfactorily performing the most complex task attempted during the analysis: discriminating between juices from the same fruit but from different brands. Moreover, the electrode “cleaning” procedure required between two subsequent potential sweeps was much simpler for the modified electrode than for the others. This electrode system was therefore shown to be a good candidate for use as an informative element in an electronic tongue applied to the analysis of other food matrices.