Learning anatomy holds specific challenges, like the appreciation of three-dimensional relationships between anatomical structures. So far, there is limited knowledge about how students construct their understanding of topographic anatomy. By understanding the processes by which students learn anatomical structures in 3D, educators will be better equipped to offer support and create successful learning situations. Using video analysis, this study investigates how students discern anatomical structures. Sixteen students at different levels of education and from different study programs were recorded audiovisually while exploring 3D digital images using a computerized visualization table. Eleven hours of recorded material were analyzed using interaction analysis and phenomenography. Seven categories were identified during data analysis, describing the qualitatively different patterns of actions that students use to make sense of anatomy: "decoding the image"; "positioning the body in space"; "purposeful seeking"; "using knowledge and experience"; "making use of and creating variation"; "aimless exploration"; and "arriving at moments of understanding". The results suggest that anatomy instruction should be organized to let the students decide how and at what pace they examine visualized images. Particularly, the discovery process of decoding and positioning the body in space supports a deep learning approach for learning anatomy using visualizations. The students' activities should be facilitated and not directed.