Human and non-human animals can use geometric information (metric information and left-right discrimination sense) to reorient themselves in an environment. The hypothesis that in so doing they rely on allocentric (map-like) representations has received wide consensus. However, theoretical models suggest that egocentric representations may represent efficient strategies for visuo-spatial navigation. Here, we provide, for the first time, evidence that a view-based strategy is effectively used by animals to reorient themselves in an array of landmarks. Domestic chicks were trained to locate a food-reward in a rectangular array of either four indistinguishable or distinctive pipes. In the key experimental series, the pipes had four openings, only one of which allowed the chicks to access the reward. The direction of the open access relative to the array was either maintained stable or it was changed throughout training. The relative position of the pipes in the array was maintained stable in both training conditions. Chicks reoriented according to configural geometry as long as the open access pointed in the same direction during training but failed when the positions of the openings was changed throughout training. When the correct pipe was characterized by a distinctive featural cue, chicks learnt to locate the reward irrespective of the stability of the direction to openings, indicating that place-navigation was dissociated from non-spatial learning. These findings provide evidence that view-based strategies to reorient by geometry could be used by animals.