The satellite thermal infrared remote sense has become a promising technique for detecting earthquake and fault activities, because of its many advantages, such as large observation area, high spatial resolution and short observation period, also owing to its good reflection to some strong earthquakes' thermal anomaly. But the thermal infrared radiation of earth surface is influenced not only by earthquake and fault activities but also by many factors such as terrain, vegetation, weather and so on. Therefore, how to extract the thermal variation really related to earthquake and fault activities from complex infrared remote sensing data is a key problem. The paper, based on several year's analyses of NOAA AVHRR data of typical tectonic active areas, gave a detail discussion to this question, put forward some applied methods for identifying and extracting thermal infrared anomaly caused by earthquakes or fault activities, and also evaluated the feasibility of these methods. The author thought the earthquake thermal infrared anomaly is a part of relative change of earth surface temperature. In order to effectively extract the earthquake anomaly and get rid of the impacts of non-earthquake factors, we need do comparative and comprehensive analyses using at least two years' satellite data in different temporal and spatial scale.