When generating 3D-landscape models using laser radar data, it is desirable to automatically extract different objects, such as buildings, trees, roads, etc, from the data set. In this study, an algorithm was implemented to extract individual trees using laser radar data. The segmentation was done in two steps. First, a segmentation based on texture measures and recording of double echoes was performed to separate man-made objects (buildings) and natural objects (vegetation). Single trees were then segmented in the identified areas of vegetation and the position, height, and crown diameter of the detected trees were estimated. The automatic recognition of the trees was compared with ground measurements and the result showed that 71 % of the trees were correcfly identified. The average positional error of the detected trees was 51.4 cm. The height and the crown diameter of the detected trees could be estimated with a standard error of 0.63m and 0.61 m, respectively.