Identification of phishing websites is very challenging task for every internet and e-mail users. To protect the information from unauthorized person classification of phishing websites is very important. In this research work , we have used many data mining based classification techniques like C4.5, SimpleCart, Random tree, SVM and MLP for classification of phishing websites with different data partitions like 75% training and 25% testing,, 80% training and 20% testing and 85% training and 15% testing. To develop a robust model , we have ensemble the models with different combinations. We have achieved better accuracy with ensemble of C4.5, SimpleCart , MLP and Random tree with all data partitions, but it achieved best accuracy as 97.16% in case of 85-15% data partition.