Semantic hyperpriming has consistently been found in normal aging. However, because the standard instructions to test semantic priming are generally ambiguous (focusing on both accuracy and speed), it is difficult to account for hyperpriming in older adults. By using the direct and mediated priming paradigms, this study investigates whether older adults' response mode at testing may explain hyperpriming. First, we show that, under identical conditions, inducing a response mode that favors speed leads to greater priming effects in older adults. The pattern of results is similar to what is observed under standard instructions. Second, prompting a response mode that favors accuracy leads to greater priming effects in younger adults. We discuss various explanations for these findings and conclude, in accordance with the Ratcliff, Thapar, Gomez, and McKoon (2004a) diffusion model, that hyperpriming in normal aging is contingent on older adults' response mode at testing.