Response selection is foundational to adaptive behavior, and considerable attention has been devoted to investigating this behavior under conditions in which the mapping between stimuli and responses is fixed. Results from prior studies implicate the left supramarginal gyrus (SMg), premotor and prefrontal cortices, as well as the cerebellum in this essential function. Yet, many goal-directed motor behaviors have multiple solutions with flexible mappings between stimuli and responses whose solutions are believed to involve prospective planning. Studies of selection under conditions of flexible mappings also reveal involvement of the left SMg, as well as bilateral premotor, superior parietal cortex (SPL) and pre-supplementary motor (pre-SMA) cortices, along with the cerebellum. This evidence is, however, limited by exclusive reliance on tasks that involve selection in the absence of overt action execution and without complete control of possible confounding effects related to differences in stimulus and response processing demands. Here, we address this limitation through use of a novel fMRI repetition suppression (FMRI-RS) paradigm. In our prime-probe design, participants select and overtly pantomime manual object rotation actions when the relationship between stimuli and responses is either flexible (experimental condition) or fixed (control condition). When trials were repeated in prime-probe pairs of the experimental condition, we detected improvements in performance accompanied by a significant suppression of blood oxygen-level dependent (BOLD) responses in: left SMg extending into and along the length of the intraparietal sulcus (IPS), right IPS, bilateral caudal superior parietal lobule (cSPL), dorsal premotor cortex (dPMC), pre-SMA, and in the lateral cerebellum. Further, region-of-interest analyses revealed interaction effects of fMRI-RS in the experimental versus control condition within left SMg and cerebellum, as well as in bilateral caudal SPL. These efficiency effects cannot be attributed to the repetition of stimulus or response processing, but instead are planning-specific and generally consistent with earlier findings from conventional fMRI investigations. We conclude that repetition-related increases in the efficiency of planning-based selection appears to be associated with parieto-cerebellar networks.