Spatial-vision research has been largely concerned with measuring and understanding the consequences of receptive-field properties measured by single-unit recording. However, in order to understand spatial-information processing in the visual system, it is equally essential to know the densities and the distribution patterns of the receptive fields. If the receptive fields are not arrayed properly across the visual field, spatial information will be lost. It has been argued, on the basis of the Whittaker-Shannon sampling theorem, that the receptors of the fovea sample the retinal image at a high enough rate to preserve essentially all the available spatial information. In this paper we show how two-dimensional sampling theory can be used to determine which combinations of receptive-field shapes and sampling patterns would preserve all spatial information from the receptors. This analysis will prove useful for determining, in conjunction with electrophysiological and anatomical evidence, what spatial information is or is not being transmitted by a given stage of the visual pathway. It may also prove useful for developing and testing theories of spatial vision.