Abstract: An approach based on maximum entropy model is proposed for event classification. This approach can classify the events by merging the features about trigger and context in event mention sentences. The key of the method is parameter estimation and feature selection, which are discussed in detail. IIS algorithm is employed for parameter estimation and incremental method is used for feature selection. Experiments are performed on management succession, meeting, terror attack, judicial adjudicate, and natural disaster in the People Daily corpus. The results show that the method can achieve much better performance than the traditional approach.