We evaluated whether two indices of light availability resolved differences among microsites within deeply shaded understories (<12% of above-canopy photosynthetic photon flux density (PPFD)) and also whether marked differences in forest canopy structure affected how the two indices related to direct measures of incident photosynthetically active radiation (PAR). Incident PAR was measured with gallium-arsenide-phosphide photodiodes at numerous points in two adjacent forest patches in Michigan, USA: one dominated by the evergreen conifer eastern hemlock (Tsuga canadensis (L.) Carrière), and the other by the deciduous hardwood sugar maple (Acer saccharum Marsh.). The two indices tested were canopy openness, measured with the LI-COR LAI-2000 plant canopy analyzer, and the percentage of above-canopy PPFD measured in the understory during overcast conditions (%PPFD). Canopy openness and %PPFD did not effectively predict the long-term mean of daily PPFD. However, both indices reliably predicted the long-term mean of daily median PPFD, an alternative standard of directly measured incident PAR that reduces the relative contribution of sunflecks. The relationships of both indices with mean daily median PPFD differed between hemlock and hardwood patches. Hence, the effect of canopy structure should be considered when using these indices to draw conclusions about differences in light availability between forest patches, particularly when narrow ranges of light availability in deep shade are important.