In order to overcome the disadvantages that the traditional k-Nearest Neighbor classification technique makes inadquate use of distribution characteristics of homegeneous data, and that it is of slow speed and low efficiency, an effective data classification algorithm based on the decision table grid is presented. The main process is to construct the corresponding decision table after discretizing training samples, to map the training samples to the corresponding grid based on the decision condition, and to map the samples to be classified to the corresponding grid, then to judge the classification of the samples by the given principle. The algorithm can quickly classify the samples to be classified and can improve the precision as well. Experiments show that it has good effect, being more suitable for high dimension data classification and capable of dealing with the training samples with many classes.