Extraction of man-made structures is one of essential issues in image processing of remotely sensed data. Due to remarkable improvements in radar sensors, high-resolution Synthetic Aperture Radar (SAR) images are available, providing detail ground surface information. In this study, a new method is developed to detect building heights automatically from a 2D GIS data and a single high-resolution TerraSAR-X (TSX) intensity image.A building in a TSX image shows layover from the actual position to the direction of the sensor due to the side-looking nature of SAR. Since the length of layover on a ground-range SAR image is proportional to the building height, it can be used to estimate the building height. To do this, we shift the building footprint obtained from 2D GIS data in the direction of the sensor. The proposed method is tested on aTSX image of Lima, Peru in the High Spot mode with a resolution of about 1 m. The result showed a reasonable level of accuracy.