Studies on "intertrial priming" have shown that in visual search experiments, the preceding trial automatically affects search performance: facilitating it when the target features repeat and giving rise to switch costs when they change--so-called (short-term) intertrial priming. These effects also occur at longer time scales: When 1 of 2 possible target colors is more frequent during an experiment block, this results in a prolonged and persistent facilitation for the color that was biased, long after the frequency bias is gone--so-called long-term priming. In this study, we explore the robustness of such long-term priming. In Experiment 1, participants were fully informed of the bias and instructed to prioritize the other unbiased color. Despite these instructions, long-term priming of the biased color persisted in this block, suggesting that guidance by long-term priming is an implicit effect. In Experiment 2, long-term priming was built up in 1 experimental session and was then assessed in a second session a week later. Long-term priming persisted across this week, emphasizing that long-term priming is truly a phenomenon of long-term memory. The results support the view that priming results from the automatic and implicit retrieval of memory traces of past trials.