The monitoring data in the power quality monitoring system increased quickly. A new method based on partial storage and selective reloading was proposed, which can solve the problem of repetitive sorting and redundant processing in the tradition methods. In the calculation of the daily index, daily data was sorted and stored partly based on saving rate. In the calculation of week (month, season or year) index, the partly saved daily data in a week (month, season or year) were merged by the multiple merge algorithm to calculate a temporary 95 percentile (CP95), which could be used to determine which daily data should be reloaded. Besides the reloaded data, all other needed data were reordered to calculate the steady index. The sorting process only needed part of the stored daily data and a small amount of reloaded data, so the redundant processing problem in traditional processing method was solved effectively. Compared with the traditional data processing method, the experimental results show the efficiency can be increased more than 3 times using the proposed method when daily sampling data is relatively small. When the number of daily sampling data is more than 2880, the efficiency can be increased more than 15 times. The larger the amount of sampling data is, the more obviously the performance improves.The method has been applied in the monitoring system of Shanxi, Hebei and other provinces successfully. It is proved in practice that the method is correct and effective.