A spatial task was used to investigate if a stimulus could set the occasion for responding to a landmark. Pigeons were trained with a positive occasion setter (OS; a colored background display) signaling the contingency between a landmark (LM; visual patterned stimulus) and the location of a rewarded response. The two most common tests of an OS (transfer tests and post-training extinction of the OS) were then conducted. In Experiment 1, two occasion setting pairs were trained (+←XA/YB→+/A−/B−) with unique spatial relationships to a reinforced goal location. Transfer tests (XB− and YA−) revealed more responding to a landmark when paired with the same OS from training (e.g., XA) than on transfer tests, which was greater still than landmark-only trials (A−). Three pigeons demonstrated good spatial control of responding by the LM on transfer tests. In Experiment 2, the contingency and spatial relationship (e.g., left or right) between LM A and the goal were signaled by the OS (+←XA/YA→+/+←ZB/C→+/A−/B−). LM C was trained without an OS to assess the role of training history during transfer. Transfer tests again indicated an OS could facilitate responding and the LM controlled the location of responding. Training history affected spatial control, but not facilitation, by LM C. Lastly, post-training extinction of X had no effect on facilitation or spatial control during subsequent XA trials. These experiments are the first to evaluate conditional control of spatial information by landmarks using both of the standard tests for occasion setting.