This dissertation was designed to study the impact of stable and unstable landmarks on repeated visual search tasks involving real-scene displays relative to no background displays in a contextual cueing procedure. Contextual cueing, a repeated visual search task, involves implicit learning through associative relationships between a set of distractors and a target location (Chun and Jiang, 1998). Stable landmarks, defined as unmoving, reliable landmarks were items like benches, road signs, and a large fountain. Unstable items, defined as capable of movement and therefore unreliable landmarks, consisted of a cat, people, and other animals. Typically contextual cueing tasks involve repeated exposures to a set of displays over several blocks in the learning phase. In a test phase, participants encounter unseen displays with random target locations. The current study involved two experiments. Experiment 1 contained real-scene displays where participants searched similar park scenes for a predetermined target amongst stable and unstable distractors. In the test phase, the stable distractors were removed in one condition, and in another, only the unstable distractors were removed. Experiment 2 was identical, although there were no real-scene backgrounds. It was predicted that participants would respond faster to the stable condition, where stable landmarks would help to predict the location of the target. The general hypothesis was that inherent stable objects such water fountains, benches, and road signs would be more informative about the location of other objects in a scene than would relative unstable objects that could move to different locations or completely out of the scene on their own such as animals or people. Overall, there was a difference in the degree to which stable and unstable objects influence performance to repeated relative to novel displays. In fact, this difference was observed for stable vs unstable objects encountered in real world scenes and those presented against a solid background. As such, the difference in the processing of stable vs unstable objects in real world scenes was reflected in a greater cost when unstable objects were removed during the test phase relative to when stable objects were removed. Overall the results indicate that participants struggle with repeated search tasks within real scenes.