In three experiments, we investigated how retrieval cues affect memory for cognitive maps. Participants first rated a list of landmarks either for the importance of the activity performed or for the frequency of visitation at each landmark (Experiments 1 and 2), or on both dimensions (Experiment 3). Landmarks ranked high and low on these dimensions were selected and served as the bases for distance estimations and route descriptions. Distances estimated using importance as a criterion for selection were significantly longer than those using frequency of visitation as a criterion. Participants in the importance group also produced more route perspective expressions than survey expressions in their descriptions, whereas participants in the frequency group did not differ in the relative use of these types of perspective expressions (Experiment 1). These findings suggest that deviations in distance estimation and the use of perspectives in route description are a result of contextual manipulation and the procedures invoked during judgments.