Abstract: In order to make the prediction of gas emission in real time and quickly,by using database and based on source prediction method,the data involved in the prediction of gas emission was divided into entity of gas fundamental parameters,entity of gas source and entity of gas emission,the station information,technical parameters influencing gas emission,and gas emission was abstracted as its entity property respectively.The EER (enhanced entity-relationship) diagram of the related entity in prediction of gas emission was built by the database modeling tool of MYSQL Workbench.District was set as the basic element for the algorithm of gas emission prediction and the algorithm was divided into four main loops：mining area,working face,tunneling face and adjacent layers.The prediction of every gas source was designed in the form of function.An accurate and expandable sharing platform was constructed based on B/S (browser/sever) model for the prediction of gas emission.