Abstract: The artificial neural network model for processing signals form ion-selective electrodes array is presented in the paper. the performance of the neural network approach was tested in the simultaneous determination of Cu²⁺/Ca²⁺ system. Its maxmium relative fitting error doesn't exceed 5.5% and maxmium predicting relative error doesn't exceed 2.4%. The results showed that the improved network is good therefore it might be used widely in processing signals from variuns ion selective electrode arrays.