Two models of regression equation (RE) and artificial neural network (ANN) are established by analyzing the relationship between the dipine drugs' molecular structure and melting points and compiling the MATLAB program. Selecting 14 dipine drugs as the modeling samples of RE and ANN methods and other 4 dipine drugs for verification, the two models are compared. And results show that the ANN model is more accurate in forecasting the drugs' melting point than the RE method.