Two experiments investigated how people use buildings and street configurations to reorient in large-scale environments. In immersive virtual environments, participants learned objects’ locations in an intersection consisting of 4 streets. The objects’ locations were specified by 2 cues: a building and/or the street configuration. During the test, participants localized objects with either or both cues. Participants were divided into a competition group and a no-competition group. The competition group learned both cues whereas the no-competition group learned the single cue for trials with single testing cue. For the trials with both testing cues, both groups learned both cues and these 2 cues were placed at the original locations or displaced relative to each other during testing. Critically, the familiarity with the environment was also manipulated: in Experiment 1, participants learned the same building at the same corner of the same intersection for all trials (familiar); in Experiment 2, participants learned different buildings at different corners of different intersections across trials (unfamiliar). The results showed that the performance in the competition group was impaired in unfamiliar environments but not in familiar environments. When displacement occurred, the participants’ preference in unfamiliar environments was determined by the response accuracy of using the 2 cues respectively, whereas participants in the familiar environment preferred the street configuration with a probability higher than what was solely determined by response accuracy based on individual cues. When the 2 cues were consistent with each other, they were combined additively in both familiar and unfamiliar environments.