There are two basic types of photoreceptors in the retina: rods and cones. Using a single stimulus viewed at two different light levels, we tested whether input from rods and input from cones are topographically segregated at subsequent levels of human visual cortex. Here we show that rod-mediated visual input produces robust activation in area MT+, and in the peripheral representations of multiple retinotopic areas. However, such activation was selectively absent in: (1) a cortical area selectively activated by colored stimuli (V8) and (2) the foveal representations of lower tier retinotopic areas. These cortical differences reflect corresponding differences in perception between scotopic and photopic conditions.