There is a long history for research of earthquake prediction, but weakness of traditional approaches to study seismic hazard have been more and more evident. Remote sensing and earth observation technology, which is a new method that can instantly acquire a large area of abnormal information caused by earthquakes, is believed to be the key to the breakthrough of the bottleneck in the study of earthquake prediction. A multi-parametric approach seems, instead, to be the most promising approach in order to increase reliability and precision of short-term seismic hazard forecast, and Thermal Infrared (TIR) anomaly is an important part of the earthquake precursors. Though many scientists have studied the correlation among TIR anomalies identified by the Robust Satellite Techniques (RST) methodology and single earthquake, there is few study to extract the TIR anomalies in long period and large study area. Moreover, a statistical analysis of TIR anomalies in relation with earthquake is needed to determine whether there is the existence of TIR anomalies before earthquake. In this paper, a refined RST data analysis and Robust Estimator of TIR Anomalies (RETIRA) index were used to extract the TIR anomalies from 2002 to 2018 in Sichuan area with use of Moderate-resolution Imaging Spectro-radiometer (MODIS) Land Surface Temperature (LST), and the earthquake catalog were also used to study the correlation between TIR anomalies and occurrences of earthquake. Most of the thermal infrared anomalies correspond to earthquakes, and statistical methods are used to prove that there is a correlation between the extracted thermal infrared anomalies and earthquakes. And this is the first time to evaluate earthquakes prediction ability with use of PPV, FDR, TPR and FNR, the statistical result shows that the prediction ability of RST in Sichuan area is limited.