In recent years,the price of Chinese herbal medicines has fluctuated violently on many occasions,with high frequency and large amplitude,which has affected the healthy development of the Chinese medicinal materials industry. Predicting the market price and price index of Chinese herbal medicines is of great significance to the smooth operation and healthy development of the Chinese herbal medicines market. The price index of 30 Chinese herbal medicines from the website of Chinese herbal medicine and the HP-LSTM-MLP hybrid forecasting model was used to predict the price index and comprehensive price index of Chinese medicinal materials,and to compare and analyze with LSTM (Long-short-term memory network),RNN (recurrent neural network),GRU (gated recurrent unit) and other predictive models. Results showed that among the two predictive effect measures of RMSE and R2score,the HP-LSTM-MLP hybrid prediction model has an RMSE of 65.33 and R2score of 0.99,which are better than that of the other three models. Finally,the generalization ability of the HP-LSTM-MLP model was analyzed. The comprehensive price index and singlevariety price index of 30 main Chinese herbal medicines in China was predicted. The average relative error of the comprehensive price index of Chinese herbal medicines was 1.89%. Among the single varieties,the average relative errors of Coptis,Forsythia and Ophiopogon was 3.36%,5.66% and 3.22%,respectively. It is indicated that the model has good generalization ability and has certain application value for the predicting the price index of Chinese herbal medicines in China.