BCI establishs a direct communication and control channel between human and computer or other electronic device. P300-based speller paradigm is an common method for BCI. In this paper, genetic algorithm and support vector machines(SVM) are used for classification of EEG. It employs Principal Component Analysis(PCA) and Fisher Discriminant Criterion to implement the feature extraction. After using PCA to reduce dimension, Fisher discriminant Criterion can further extract effective features and improve the accuracy of classfication. This paper employs SVM to classify electroencephalogram.