Bone segmentation in radiographic imaging is an intermediate level processing stage for an automated vision system for the skeletal assessment of children. It is one of the challenging problems in medical image analysis due to high noise levels, and low contrast with non-uniform and complex intensity distribution of radiographic image. In this paper, we present an algorithm for automatically segmenting bone from hand radiograph using watershed transform (WT) with multistage merging. WT and image preprocessing are first utilized for initializing the primitive (homogeneous) regions. Next, the merging process with two stages is applied to generate the final segmentation result. The first merging stage is performed to remove unnecessary regions between tissue and background in order to enhance the accuracy of the merging. The second stage is used to detect bones in X-Ray image. A merging criteria based on the variation of intensities along the common boundary between the pair of adjacent regions is utilized to control the merging process. Experiments are carried out on X-ray images of hand bone. The experimental results show that through the combination of WT and multistage merging, an accurate and robust segmentation can be achieved .