In each of five experiments Ss made direct assessments of interpoint distances within a stimulus configuration. A multidimensional scaling method was applied to make explicit the form of the subjective configuration. The stimulus configurations consisted of small light points arranged in a two- or three-dimensional array in a dark room, and of real stars in the sky. All the data were adequately accounted for by a configuration constructed in Euclidean space of the appropriate dimensionality. That was true even in the situation where alley experiments with the same Ss gave the result that is usually regarded as evidence for a hyperbolic binocular space. The Euclidean interpretation entails a more complicated form of correspondence between physical and visual spaces than the hyperbolic interpretation.