Rule‐governed behavior and derived stimulus relations have always shared strong conceptual links within behavior analysis. However, experimental analysis linking the two domains remains limited. The current study consisted of three experiments that aimed to continue to bridge this experimental gap. The first experiment sought to establish the extent to which a training version of the implicit relational assessment procedure (IRAP) could be used to establish and successfully reverse experimentally established derived relations. The results suggested that the Training IRAP could successfully produce derived reversals. Experiments 2 and 3 explored the extent to which reversed derived relations would control rule‐governed behavior when the contingencies for rule‐following competed with the rule. In Experiment 2, the task contingencies were immediately in opposition to the (reversed) derived rule, and participants generally responded in accordance with the task contingencies, rather than the rule. In Experiment 3, the task contingencies were initially rule‐consistent before a contingency reversal that later made them rule‐inconsistent. Here evidence of rule‐persistence emerged. The results of the research are considered within the context of a recent framework that has emerged out of RFT for analyzing the dynamics involved in derived relational responding.