Abstract: Three-layer artificial neural network(ANN)models with back propagation of error have been used to the analysis of chlorpromazine tablets．The networks architecture and parameters were optimized．The network gave an average relative error of 1.22％ by using f(X)=1/(1+e^-x)as the input-output transformation function and setting the number of hidden nodes to 15．The network was used to determine the content of chlorpromazine in pharmaceutical preparations．The results were in fair agreement with those obtained by drug standard method and the PLS methods．