Since first presented by Shepard and Metzler, Science 1971, 171: 701–703, mental rotation has been described as a rotary transformation of a visual stimulus allowing it to be represented in a new orientation. For a given stimulus, the transformation is thought to occur at a constant speed, though speed may vary between stimuli; three‐dimensional abstract shapes made out of blocks tend to be rotated much more slowly than alphanumeric characters or line drawings of common objects. Rotation is also presumed to be performed through the shortest angle. These assumptions are based upon the fact that response times tend to increase with angle of rotation, peaking at 180° of separation for abstract block figures or from upright for common objects and alphanumeric stimuli. The symmetry about 180° provides evidence supporting rotation through the shortest angle. In order to determine the shortest direction, the current orientation of the stimulus is assumed to be known prior to mental rotation. Moreover, in order to determine the current orientation of a common object or alphanumeric stimulus, it is assumed the stimulus is identified prior to mental rotation because the current orientation is defined by what the object is. In mirror/normal discriminations or left/right facing discriminations of rotated stimuli response times are often examined by collapsing over response options as this variable is assumed to be uninteresting in terms of mental rotation. This article examines these assumptions, and suggests that many of them are not entirely safe. WIREs Cogn Sci 2017, 8:e1443. doi: 10.1002/wcs.1443This article is categorized under: Psychology > Theory and Methods