Abstract: The prediction errors are statistically analyzed for the method of the ionospheric total electron content(TEC) short-term forecasting in China region. The results show that the prediction error rate gradually becomes larger from north to south China, and have the larger value in vernal and autumnal equinoxes. The diurnal variation of prediction errors has a single-peak at about 15:00. While the relative errors have the largest values in winter and the smallest values in summer, and exist two obvious peaks at around 8:00 and 20:00. Overall, the mean of the relative errors is about 12% with 1 hour in advance. At the same time, it can be found that the prediction error for the single station forecasting has larger proportion in the whole method by comparing it to the reconstruction error. Therefore, to improve the single station prediction is such an effective way to improve the accuracy of the whole method. In the meantime, it is necessary to increase and well distribute the GPS receiving stations.