The study of underground coal fires in China began in the 1960s. The huge loss of coal resources and the ecological disasters caused by coal fires in northern China promoted the study of these fires. Various remote-sensing methods are used to detect ground anomalies due to underground fires. However, locating these fires using remote-sensing data is a difficult task. Ground thermal anomalies are useful for locating underground coal fires. Thermal-geological models link ground thermal anomalies to underground fires. A method of point-source inversion is applicable to a simplified model for the inverse locations of underground coal fires. When tested with data from the Wuda area in the inner Mongolia Autonomous Region of China, this method exhibits encouraging results.