Mineral waters have different tastes, according to an electronic tongue (E.T.), constructed by 29 non selective ion sensors. High input impedance and low noise op-amps have been employed for the voltage amplification of each sensing point and the multicomponent analysis technique has been utilized for the data processing of chemical images and their representation. Well separated chemical images of seven different mineral waters have been obtained and this indicates the suitability of the designed E.T. to distinguish liquids with different chemical composition. Preliminary data related to images of polluted mineral waters are also presented and commented.