Abstract: An incremental support vector data description (ISVDD) method is proposed for online learning on SVDD. It is a iterative process for training the incremental sample where each iteration composes of two steps: (1) ISVDD can fully use prior training model information to divide the existent samples and incremental sample to three sets, and compute the property value for each sample. (2) Under six conditions, the migrating of sample can be achieved, which brings in the coefficient change values for samples. Compared with standard SVDD, the proposed method reduces training time for online incremental samples，releases large memory burthen and can reach the classification accuracy as same as that of SVDD. The experimental results prove the efficiency and validity of the proposed method.