This chapter presents an unsupervised parametric mixture model for automatic three-dimensional lung segmentation. Development of a robust lung segmentation algorithm aims to aid in the early detection of lung diseases, most notably lung cancer. If diagnosed early, lung cancer patients can survive to up to a year. However, there usually are no clear symptoms of lung cancer. Segmentation of the lung tissue would allow for early diagnosis of such symptoms. Hence, the proposed algorithm aims to reduce mortality as a significant stage in a computer-aided diagnosis system. Furthermore, in this chapter, we briefly describe the anatomy and key functions of the lung and overview related work on lung segmentation. Finally, we discuss our experimental methods with computed tomographic images.