This paper presents a set of Augmented Reality (AR) based interaction techniques for spatial analysis of medical datasets. Computer-aided medical planning tools such as our Virtual Liver Surgery Planning System require precise and intuitive interaction for the quantitative inspection of anatomical and pathological structures. We argue that AR is a superior tool compared to desktop 2D or 3D visualization for performing such analysis, because it allows true direct manipulation of 3D virtual objects in space, while rendering the medical data in the familiar context of the user’s own body.