A sample and class incremental learning algorithm based on hyper ellipsoidal is proposed. For every class, the smallest hyper ellipsoidal that surrounds most samples of the class is structured, which can divide the class samples from others. In the process of incremental learning, only the hyper ellipsoidal of every new class is trained and the history hyper ellipsoidals that increment new samples are retrained. For the sample to be classified, its class be confirmed by the hyper ellipsoidal that surrounds it. If the sample is not surrounded by all hyper ellipsoidals, the membership is used to confirmed its class. The experiments are done on Reuters 21578, and the experiment results show that the algorithm has a higher performance on classification speed and classification precision compare with hyper sphere algorithm.