Cast shadows provide a strong cue to depth (Mammassian, Knill, & Kersten, 1998). Search times for inconsistent shadows differ when images are presented upside-down, compared to upright. Some studies have shown faster detection in upside-down images (Rensink & Cavanagh, 2004), while others have found faster detection for upright images with large shadow discrepancies, but the opposite pattern for small discrepancies (Lovell et al., 2009). Lovell et al. suggest that this pattern of results is explained by a coarsely scaled shadow processing mechanism that only comes into play with light-from-above stimuli. Here we report a series of experiments that explore whether shadows are coarsely processed. Stimuli feature floating discs casting a shadow onto a vertical (fronto-parallel) surface. The observer was asked to identify which disc was located furthest towards the observer; the only available cue was the cast-shadow. Stimuli contained a cue-conflict, where higher and lower spatial frequency components convey different physical depths. By examining the location of the point of subjective equality relative to the two cuedepths, we can estimate perceptual weightings for the coarse and fine-scaled information, we can also examine the differences in these weightings for upright and upside-down images. In upright images coarser cues seem to receive stronger perceptual weighting.