Four experiments examined the roles of extended surfaces and the number of points in the boundary superiority effect in goal localization. Participants learned the locations of 4 objects in the presence of a boundary, landmarks, or both in an immersive virtual environment by reproducing the locations with feedback. Participants then localized the objects in the presence of either the boundary or the landmarks during testing without feedback. The results showed that when both 1 landmark and a circular boundary were presented during learning, localization error during testing increased significantly when only the landmark was presented during testing, whereas localization error did not increase when only the boundary was presented during testing, thus demonstrating a boundary superiority effect. This boundary superiority effect was not observed when 36 landmarks forming a circle and a circular boundary were presented during learning. The landmark superiority effect was observed when 36 landmarks, forming a circular shape, and 1/36th part of the circular boundary were presented during learning. Furthermore, when a varied number of landmarks were presented with a circular boundary during learning, the localization error when the boundary was removed during testing was negatively correlated with the number of the landmarks. These results indicate that the superiority of a circular boundary to a landmark might be due to the larger number of points in the circular boundary but not due to the extended surface of the circular boundary.