In the present paper, the simultaneous determination of multi-components in Chinese herbal medicine was performed by artificial neural network-UV spectrometry. The interference of other components was eliminated by self-revising and self-simulation of the virtual component. The double ANN including training and simulation network was established, and the capabilities of self-recognition and self-studying were improved. Therefore, the prediction accuracy of multicomponent content was improved greatly in the complicated Chinese herbal medicine system. The contents of aesculin and aesculetin, which were extracted from 21 Cortex Fraxinis, were predicted. Comparing the results with those of HPLC, the prediction accuracy was more than 90% within the relative errors less than 10%. The measurement precisions of aesculin and aesculetin were 0. 37% and 1. 5% respectively.