Previous studies show that following disorientation children use the geometry of an enclosed space to locate an object hidden in one of the corners [e.g. (Harmer, L., & Spelke, E. (1996). Modularity and development: A case of spatial reorientation. "Cognition, 61," 195-232)]. These studies have used a disorientation procedure that involves rotating the viewer (with eyes closed). Here, we examine 18- to 25-month-olds' spatial coding in two disorientation tasks--involving either "viewer" or "space" rotation. Importantly, the rotational movements in both tasks could not be visually tracked. Children were tested in either task (viewer- or space-movement) from either inside or outside a triangular (isosceles) space (with one unique and two equivalent corners). In the viewer-movement task, performance was above chance, regardless of which corner contained the object. In the space-movement task, performance was above chance at only the unique corner. On both tasks, performance was better from inside the space than from outside. The implications for how children determine location are discussed.