Urea and creatinine biosensors based on urease and creatinine deiminase, respectively, covalently immobilized onto ammonium selective electrodes, were included in an array together with sensors sensitive to ammonium, potassium and sodium. Generic sensors to alkaline ions were also included. All the sensors used were of all-solid-state type, employing polymeric membranes and having rather nonspecific response characteristics. A response model based on artificial neural networks was built and tested for the simultaneous determination of urea, creatinine, ammonium, potassium and sodium. The results show that it is possible to obtain a good multivariate calibration model. In this way, the developed bioelectronic tongue was successfully applied to multidetermination of the five species in raw and spiked urine samples. Predicted concentrations showed a good agreement with reference methods of analysis, allowing a simple direct method for determining urea and creatinine in real samples. At the same time, this method permitted to obtain the concentrations of the alkaline interferences (endogenous ammonium, potassium and sodium) without the need of eliminating them.