<p>It has been proposed that the control mechanism underlying the congruent sequence effect (CSE) resolves spatial conflict by suppressing an automatic link between spatial stimulus and response codes. However, previous explanation for the inhibitory influence on the automatic link was based solely on demonstrating that the control mechanisms are specific to the task-irrelevant stimulus and response dimensions, rather than the automatic link itself. The present study investigated whether the occurrence of the cross-task CSE depends on the sharing of an automatic link between two tasks, when the two tasks share an irrelevant stimulus dimension and response mode. In the experiment, participants performed alternating trials of a horizontal Simon task and a vertical spatial Stroop task. The cross-task CSE was significant only when response keys for each task were spatially configured to correspond to the axes of possible target stimulus locations (H-V configuration of response sets), making the two tasks share an automatic link. However, no cross-task CSE was obtained when all response keys were aligned horizontally (H–H configuration of response sets). With the H–H configuration, there was no spatial correspondence between the stimulus and response locations in the vertical spatial Stroop task, whereas the horizontal task involved such a correspondence, preventing the two tasks from sharing an automatic link. These results indicate that sharing of an automatic link between spatial conflict tasks is crucial for the occurrence of the cross-task CSE.</p>