Spatial information can incidentally guide attention to the likely location of a target. This contextual cueing was even observed if only the relative configuration, but not the individual locations of distractor items were repeated or vice versa (Jiang & Wagner in Perception & Psychophysics, 66(3), 454-463, 2004). The present study investigated the contribution of global configuration and individual spatial location to contextual cueing. Participants repeatedly searched 12 visual search displays in a learning session. In a subsequent transfer session, there were four conditions: fully repeated configurations (same as the displays in the learning session), recombined configurations from two learned configurations with the same target location (preserving distractor locations but not configuration), rotated configurations (preserving configuration but not distractor locations), and new configurations. We could show that contextual cueing occurred if only distractor locations or relative configuration, randomly intermixed, was preserved in a single experiment. Beyond replicating the results of Jiang and Wagner, we made an adjustment to a particular type of transformation - that may have occurred in separate experiments - unlikely. Moreover, contextual cueing in rotated configurations showed that repeated configurations can serve as context cues even without preserved azimuth.