Shoulder fractures have been classified for over 3500 years (Chaps. 4 and 6). Various classifications have been proposed over time according to changes in diagnostic tools, anatomical knowledge, therapeutic options, and intellectual interests. The value and purpose of a classification system are interdependent, and the desired properties may conflict, making the perfect classification an unattainable idea. The classification user learns to recognize and order similarities of fracture characteristics according to the prevalent way of clinical thinking. In this chapter, I will discuss the desirable properties of a classification of shoulder fractures. I will briefly touch on the philosophical discussion of morbid phenomena as natural kinds and the ways in which human cognition limits the possibility of determining true classification categories. I conclude that the classification dynamically changes in accordance with the development of scientific tools and methods and the sociocultural context.