Forecasting the weather diseases is a complex process with strong uncertainty. In order to improve the forecast accuracy, the paper proposed a sort of forecast model for weather diseases based on the artificial neural network. By means of ANN, lots of sample data could be trained so as to optimize the connecting weighting value of ANN, finally the forecast effect of weather diseases has been intuitively verified by experiment simulation. The factor resulted in weather diseases could be the following weather parameters, such as temperature of a day, relative humidity, air pressure, wind speed and so on. The paper takes respiratory disease treatment records as an example, the simulation result points out that it is consistent between the change trend of being sick and the actual situation. The study result shows that the model is reasonable and feasible, and could forecast weather disease more effectively.