Six experiments tested the assumption that the mental rotation process is purely inserted into a mirror-normal discrimination task. In Exp 1, Ss took significantly longer to respond to upright characters in blocks containing rotated stimuli than in blocks containing only upright stimuli. Exps 2 and 3 showed that this rotational uncertainty effect was not caused by the need to determine stimulus orientation, and Exp 4 showed that it was independent of the visual quality of the stimulus. Exp 5 showed that the effect was greatly reduced when Ss performed a goo-go task rather than choice reaction time (RT), and Exp 6 showed that it was independent of the complexity of the response required in the choice task. Results suggest that response selection in a choice RT mirror-normal discrimination task is altered when mental rotation is added, violating the assumption of pure insertion.