The present paper is concerned with the representation of spatial location in young children. We report six experiments which indicate that the basic framework for coding location is present early in life. Later development consists of an increasing ability to impose organization on a broad range of bounded spaces. In the first four experiments, we examined whether very young children, like adults, can locate objects in a homogeneous space, estimating by eye the location of those objects within some frame of reference. Results show that children from 16 to 24 months are able to use distance to code the location of an object hidden in a large sandbox. Coding of distance is not dependent on a juxtaposed outside landmark, nor on the child's own position. In the last two experiments, we examine whether young children, like adults, code the location of an object hierarchically--not only as being in a particular location in a bounded space, but also as being within a larger segment of that space. The pattern of bias in responding provides evidence for such two-level coding of location. The age at which children impose subdivisions on a space depends on the nature of that space. The sandbox is subdivided by 10-year-olds, but not by 4- or 6-year-olds. In contrast, a rectangle of similar shape drawn on paper is subdivided even by 4-year-olds. We argue that 16-month-olds in the sandbox studies also use hierarchical coding, treating the whole box as a category, although they do not divide it into subsections.