Although high-fidelity digital models of human anatomy based on actual cross-sectional images of the human body have been developed, reports on the use of physical models in anatomy teaching continue to appear. This article aims to examine the common features shared by these physical models and analyze their educational value based on the literature on cognition, learning, and external representations. A literature search on these physical models in three popular anatomy journals published over a 10-year period from 2001 to 2010 found that all of them have low fidelity: they oftentimes do not closely resemble the regions of the human body they are representing. They include only a small number of the structures that exist in these regions of the human body and do not accurately represent the shape and surface details of these structures. However, these models strongly correspond to the human body in the spatial relationship of the represented structures, which is crucial to achieving their educational purpose of teaching three-dimensional comprehension and anatomical reasoning. The educational value of these models includes acting as memory aids, reducing cognitive overload, facilitating problem solving, and arousing students' enthusiasm and participation. Because these models often lack a close resemblance to the human body, their use in anatomy teaching should always be accompanied by adequate explanations to the students to establish the correspondence between the models and the parts of the human body they are representing. (Contains 1 table and 1 figure.)