Maljkovic and Nakayama (1994) found that pop-out search performance is more efficient when a singleton target feature repeats rather than switches from 1 trial to the next--an effect known as priming of pop-out (PoP). They also reported findings indicating that the PoP effect is strongly automatic, as it was unaffected by knowledge of the upcoming target color. In the present study, we examined the impact of visual imagery on the PoP effect. Participants were instructed to imagine a target color that was opposite that of the preceding trial (e.g., if the prior target was red, then imagine green). Under these conditions, responses were faster for targets that matched the imagined color than for targets that matched the previous target color, reversing the typical PoP effect. There was no such reversal of the PoP effect for participants asked to verbalize rather than imagine an upcoming target color. In Experiment 3, we explored whether the PoP effect was indeed eliminated in the prior experiments, or instead obscured by the opposing visual imagery effect. Two conditions were compared, 1 in which a PoP effect could oppose the visual imagery effect, and another in which no such effect was possible, allowing inferences about whether a PoP effect was present. The results indicated that the PoP effect was present, but obscured by the larger visual imagery strategy effect that pushed performance in the opposite direction. Overall, the results suggest that the PoP effect is sensitive to top-down strategies that involve visual representations.