Pointing gestures are a cornerstone of human communication. However, to use them effectively requires some understanding of how accurate pointing is. One hypothesis is that humans represent the accuracy of their pointing gestures by means of a pointing cone. This cone is thought to protrude from the pointing finger and encompasses all objects that might be considered pointed-at. Here, we tested this hypothesis by asking participants to walk toward and point at one of several cities on fictive maps. We expected that participants walked as close to the map, as was necessary to make the pointing gesture unambiguous. If participants relied on a pointing cone to choose from which distance to point, we expected the chosen pointer-to-map-distance to depend linearly on the distance between the referent and adjacent cities. This expectation was borne out in a series of experiments. However, participants went much closer to the map than would be expected from the pointing-cone hypothesis. The data suggest that while a pointing-cone like representation might be involved in choosing how we employ pointing gestures, other factors seem to play a far more important role.