Abstract: The underground fluid observational data of Dian 02 well at Yuanmou seismic station from 2008 to 2024 were selected, and four types of interference were identified by comparison with normal observational data: technical system failure, site environmental, human factor, and other accidental interference. The prediction efficiency of the underground fluid observational data from Yuanmou seismic station was analyzed. The prediction efficiency R values of all underground fluid observational data of Yuanmou seismic station were calculated using the threshold, rate difference, trend turning, and annual change detection methods. The results showed that the water temperature, water level, and gas radon R values calculated using these four methods were all greater than 0, indicating that they have predictive significance. The prediction efficiency of the water temperature calculated using the rate difference method was R greater than R₀, indicating predictive significance. The gaseous mercury R values calculated using the four methods were all less than 0, indicating that gaseous mercury had no effective predictive significance.