How close do 3d computer graphics systems come to representing real visual scenes? Often graphics represnetations appear to lack the fine detail which is a common feature of the visual environment. In addition, the luminance adn dynamic range of a VDU screen are typically lower than those of natural scenes. Does this matter? Existing measures of the fidelity of photorealistic graphics take account of the Fourier content of the image, but this may not reveal specific errors in the portrayal of lighting and shadows; furthermore, it could be argued that one needs a measure which involves assessment of performance on a specific visual task - performance on that task in a real scene is compared with performance in the graphics scene to give an index of fidelity. Perception of the lightness of patches varying in reflectance may be a suitable candidate for the choice of visual task. It is simple to perform, and it is known that lightness contancy depends on the successful perception of lighting and the 3D structure of a scene. We have compared lightness perception in a real scene with similar perception in a ray-traced graphics representation. Fifteen subjects wer required to match reflectance of various gray patches to a reference set. For simple scenes, there was a high degree of correlation between the real and graphics results, suggesting that the graphics scenes were perceptually equivalent to the real scene. For scenes of greater complexity, this correlation may be lowered, suggesting imperfections in the graphics procedure.