The growth and properties of Bi-YIG magnetooptic film materials are analysed,the artificial neural network(ANN) model—Back-Propagation(BP) network model,which is used to predict the bismuth content in Bi-YIG films,was established.An algorithm of error-condiction Back-Propagation(ECBP) is presented based on the BP algorithm.It is proved that the model and the algorithm could successfully predict the bismuth content.It is also a basis for further materials design.