We investigated the use of landmarks by capuchins to solve spatial search tasks. In experiment 1 one subject learned to find a hidden reward in the middle of a 4-landmark configuration. During probe trials, with the landmark configuration expanded and no reward, the capuchin mainly searched near two of the four landmarks thus showing it used the landmarks as beacons. In experiment 2 two subjects learned to find a reward halfway between two landmarks, with the between-landmark line variously oriented with respect to the room. During probe trials, with the landmark configuration expanded and no reward, the capuchins no longer searched in the middle of the landmark configuration. The capuchins searched between the landmarks, but at the training distance from each landmark separately. To do so, the capuchins may have memorized a certain distance to cover, beginning from a landmark, or exploited different types of perceptual information. Therefore, the capuchins use nearby landmarks to locate a goal, but not configurationally. The results are compared with those of previous studies with other animal species and discussed in relation to issues of spatial cognition