Reflection densitometry has been widely used to measure the density difference of the bleachable cone photopigments in human eyes. Most such measurements make a series of assumptions concerning the amount of scattered light to derive an estimate of the true cone photopigment density from the density difference measurements. The current study made three types of measurements of the light returning from the eye before and after bleaching: the amount of light returning in the “directed” reflection, which is a double-pass estimate of the cone photopigment density; the amount of light in undirected or diffuse reflection; and the amount of fluorescence from lipofuscin in the RPE, which provides a single-pass measurement of optical density difference. For a 1 deg foveally fixated field, the density difference estimates for the three measurements were 0.68, 0.21, and 0.22 respectively. The lipofuscin fluorescence was found to be unguided. The background density difference was non-negligible and very close to the single pass estimate from fluorescence. These measurements each involve potentially different pathways of light through the retina, and therefore place different constraints on models of these pathways. A simple model comparing the directional and the fluorescence optical densities produced retinal coverage estimates around 70-75%. Estimates of the shape factor of the single pass optical Stiles-Crawford effect were evaluated from the dark-adapted and bleached fluorescence measurements. The values were closer to those obtained from psychophysical methods than to the double pass optical Stiles-Crawford shape factors obtained directly from retinal reflectometry.