The melt blowing nonwoven fabrics are characterized by high porosity, tiny pore diameter and ultrafine fibers, which make them well serve the function of high efficiency filter materials used in various fields. The filtration properties of melt blowing nonwovens are affected by the pore structure of nonwovens which is strongly related to the processing parameters. However, it is difficult to establish physical models on the relationship between the processing parameters and air filtration properties. In this research, two modeling methods are used to predict the air filtration properties. Due to their excellent abilities of nonlinear mapping and self-adaptation, the artificial neural network model provides an alternative to conventional methods. The results reveal that the prediction of artificial neural network model is better than the linear multiple regression model.