This study deals with methods for quantification of automobile traffic information from thermal images remotely sensed by means of a infrared thermal camera. The images of a crossing of principal roads were acquired at one-hour intervals on clear summer and winter days. Upon digital image processing, it was clear that surface temperature of roads varies with the length of time traffic is stationary. The surface temperature of roads was calculated in various values of ratio of the stationary time, and the ratio was estimated by superimposing the calculated values on the measured one. As a result, the estimated value was in good agreement with the one obtained .from videotape which had recorded automobile traffic conditions. In addition, a supervised method for extracting automobiles on a road from thermal images was presented.