Translation accounts have argued that the presence of a Stroop effect in the context of a nonvocal untranslated response is caused by verbal mediation. In its simplest form, color-labeled buttons are translated into a verbal code that interferes with color responses. On this logic, in the reverse Stroop task (identify the word; ignore the color), responses made via word-labeled buttons should also be verbally mediated. Thus, no reverse Stroop effect (RSE) should be seen. The authors tested this verbal mediation hypothesis in 4 reverse Stroop task experiments. An RSE was observed across 4 experiments. The results of Experiments 3 and 4 suggest that this RSE is driven by response competition. It is argued that the data from these 4 experiments are inconsistent with unadorned translation accounts of the RSE but consistent with an account in which the strength of association between a stimulus and a specific response plays a major role.