We report a series of experiments in which spatial judgments of the real world were compared with equivalent judgments of photographs of the real-world scenes. In experiment 1, subjects judged the angle from the horizontal of natural slopes. Judgments of slope correlated with true slope (r=0.88) but judgments were in general overestimates. Equivalent judgments of slope in photographs again correlated with true slope (r=0.91) but judgments tended to be overestimates for small angles (6°) and underestimates for larger angles (up to 25°). In experiment 2 slope judgments were made under laboratory conditions rather than in the natural world. The slopes, which were viewed monocularly, varied from 5° – 45°, and were either plain, or textured, or included perspective information (a rectangle drawn on the surface) or had both texture and perspective. Judgments were overestimates, but the correlation with true slope was high (r=0.97). Slopes with either texture or perspective were judged more accurately than plain slopes, but combining texture and perspective information conferred no further benefit. Judgment of the angle of the same slopes in photographs produced similar results, but the degree of overestimation (closer to the vertical) was greater than for the real slopes. In experiment 3, subjects either judged the distance of landmarks ranging from 200 m to 5000 m from the observation point, or judged distance to the landmarks in photographs. In both cases subjects' judgments were well described by a power function with exponents close to one. Although there are large individual differences, subjects' judgments of slope and distance are accurate to a scale factor, and photographs yield similar judgments to real scenes.