Abstract: In order to further understand the seismic prediction efficiency of abnormal trend change of underground fluid data in Liaoning area, the matlab program is compiled according to the principle of fluid discipline trend turning method, and the trend change of daily mean observation data of underground fluid water temperature, water level and water radon in Liaoning area is analyzed, and the R value scoring method is used to test the seismic prediction efficiency. The test results show that the Qipanshan water temperature, the Dashiqiao Zhoujia dynamic water level, the Anshan well static water level, the Dandong Tangchi static water level, the Panjin Qiyi well static water level and the Xiuyan 1 well dissolved gas radon concentration are six. The R value is used to check the R value when R > 0, which is of predictive significance. Among them, the water temperature of Qipan Mountain and the R value of Zhoujia dynamic water level of Dashiqiao passed the significance test of 97.5 % confidence level, which had significant predictive significance. These six items have similar regularity in the trend turning shape and turning amplitude, and are concentrated in the Tanlu fault zone and its branch structures in the southern Liaoning and the Bohai Rim.This law has practical reference value for earthquake prediction in Liaoning in the future.