For accurate predictions of shallow landslide occurrences, it is essential to develop techniques which can efficiently detect locations of rainwater convergence within natural hillslopes. As one of such techniques, this study evaluated applicability of thermal infrared remote sensing. We studied a hillslope covered with a plantation of40�]year�]old Japanese cedars, which showed relatively flat and homogeneous canopy surfaces. A thermal image of the forest canopy was taken during hot and dry daytime, by using an infrared thermography from a location100�\250m distant from the studied hillslope. The image clearly detected some areas which showed lower canopy surface temperatures than surrounding areas. Along a line transecting both high�]and low�]canopy temperature areas, distributions of soil thickness and soil water content were measured by using a combined�]penetrometer�]moisture probe, and an electrical resistivity survey, which employed a dipole�]dipole electrode arrangement, was conducted. As a result, the lower canopy temperature area showed higher soil water content and lower electrical resistivity than the high canopy temperature area, indicating that the thermal infrared remote sensing can be effectively used to detect locations of rainwater convergence within hillslopes, and, consequently, can be an efficient technique for locating slopes with high landslide�]vulnerability. The result was supported by occurrences of two small landslides within the low canopy temperature area.