We administered a stop-change paradigm, an extended version of the stop task that requires (a) stopping an ongoing motor response and (b) changing to an alternative (change) response. Performance of a group of patients diagnosed with Parkinson’s disease (PD) and taking dopaminergic medication was compared with that of matched healthy control (HC) participants. Behavioral results indicated that response latencies to the initial go signal did not distinguish between the 2 groups, but that stopping latencies were prolonged in PD patients. In addition, the change response was delayed in the clinical group, indicating difficulties in flexibly changing to alternative motor actions upon external cues. The change deficit in PD related to the inhibition deficit. This dependence points to a serial processing architecture in PD according to which the stopping process has to finish before the change process can be initiated. In contrast, the HC group showed parallel stop and change processing. Analyses of sequential trial effects suggest that both HC and PD patients are susceptible to aftereffects of action override, due to the consequences of the automatic retrieval of recent associations between action and goal representations. Interestingly, postchange performance of the clinical group was hampered disproportionately, suggesting that PD is associated with an impairment in overriding previously formed action-goal associations. These findings support the notion that both higher-order cognitive control processes, such as inhibiting and changing actions, as well as lower-order feature binding mechanisms rely on basal ganglia functioning and are compromised by the basal ganglia dysfunction caused by PD.