In contrast to traditional conflict paradigms, which measure interference from (over)trained associations, recent paradigms have been introduced that investigate automatic interference from newly instructed, but never executed, associations. In these prospective-instruction paradigms, participants receive new task instructions (e.g., if cat press left, if dog press right), but before they have to apply the instructions, they are first presented with another task that measures the automatic interference from the instructed task information. The resulting instruction-based congruency (IBC) effect is assumed to reflect the strength with which instructions are encoded and maintained in view of their future application. If this assumption holds true, the IBC effect should be inversely related to the speed with which the task instructions are eventually executed. To test this hypothesis, we administered a prospective-instruction paradigm to a large sample of 184 participants and observed a negative correlation between the IBC effect and mean reaction time on the instructed task. Similarly, an analysis looking at within-subject variations in the IBC effect and instructed task reaction times showed the same negative relation. Finally, we also present additional analyses suggesting this effect is independent from standard (experience-based) interference effects, and report explorative analyses that tested possible correlations with personality trait questionaires. Together, these findings confirm a key assumption of the IBC effect in prospective-instruction paradigms, and further support the use of this paradigm in instruction research.