Computer Tomography (CT) is one of the most used imaging techniques for the proper visualization of body structures with an appropriate resolution for diagnosis and medical procedures. Regarding cancer treatment, CT images are used to guide radiotherapy offering an effective treatment while reducing radiation exposure. In this paper, an automatic mandible segmentation algorithm is proposed. 25 CT volumes were obtained from 2015 MICCAI challenge: Head and neck Auto Segmentation Challenge database. At first, preprocessing is performed by an image size homogenization, followed by a histogram specification and grayscale morphology operations. After that, a vector quantization is done to obtain a proper static threshold for image binarization. Finally, a 3D template matching using cross-correlation was performed to find a point inside the mandible followed by a geodesic dilation using the binary image as mask. For this structure, a mean Dice index of 0.80486 was obtained for the optional test dataset of the challenge.