Studies examining possible priming effects on visual search have generally shown that repeating the same type of search facilitates or speeds performance. But such studies typically assess any priming via measuring response latency, in tasks where accuracy is at or near ceiling. This leaves open the possibility that criterion shifts alone might produce the apparent improvements, and such shifts could plausibly arise when, say, a particular type of repeated search display becomes predictable. Here we assessed criterion free perceptual sensitivity (d') for visual search, in two experiments that used brief masked displays to bring performance off ceiling. In experiment 1, sensitivity for conjunction search improved with successive repetitions of the same type of search, with sensitivity enhanced for both target-present and target-absent trials. In experiment 2, sensitivity for a search task requiring discrimination on a color-singleton target likewise showed enhancement with repetition. We conclude that priming in visual search, arising due to repetition streaks, is characterized by genuine improvements in perceptual sensitivity, not just criterion shifts.