This study investigated how young children's increasingly flexible use of spatial reference frames enables accurate search for hidden objects by using a task that 3-year-olds have been shown to perform with great accuracy and 2-year-olds have been shown to perform inaccurately. Children watched as an object was rolled down a ramp, behind a panel of doors, and stopped at a barrier visible above the doors. In two experiments, we gave 2 and 2.5-year-olds a strong reference frame by increasing the relative salience and stability of the barrier. In Experiment 1, 2.5-year-olds performed at above-chance levels with the more salient barrier. In Experiment 2, we highlighted the stability of the barrier (or ramp) by maximizing the spatial extent of each reference frame across the first four training trials. Children given a stable barrier (and moving ramp) during these initial trials performed at above-chance levels, and significantly better than children given a stable ramp (and moving barrier). This work highlights that factors central to spatial cognition and motor planning-aligning egocentric and object-centered reference frames-play a role in the ramp task during this transitional phase in development.Recent studies have revealed something quite striking: older children sometimes fail to show competencies that infants have been thought to have. For example, researchers presented 2-and 3-year-old children with a ball that rolled down a ramp (Berthier, DeBlois, Poirier, Novak, & Clifton, 2000). The ball went behind an occluder with four doors and stopped at a barrier, visible above the occluder. The barrier could be placed beside any of the doors. Berthier and colleagues (2000) found that 2-and 2.5-year-old children were not able to select the correct door when asked to find the ball in this "ramp task." In fact, out of 16 2.5-year-olds, only 3 performed at above-chance levels. Three-year-olds were able to reliably choose the correct door at above-chance levels. This is surprising because infants in a related violation-of-expectancy task appear to understand that solid objects behind an occluder stop at solid barriers (see Spelke, Breinlinger, Macomber, Jacobson, 1992). We are left to wonder how can infants demonstrate an understanding of hidden objects and solid barriers while toddlers cannot.