Seedling recruitment in temperate oaks is expected to depend mostly on biotic filters, given the relative independence from resource limitation conferred by large seed size. We investigated the magnitude and fine-scale spatial pattern of red oak (Quercus rubra) recruitment in a mature forest in southern Québec. We compared the scale of operation of experimentally assessed seed and seedling predation to that of several abiotic factors. We sought to interpret concordance or discordance between seed rain and seedling recruitment in terms of spatial heterogeneity in these environmental filters. Most of the demographic variables (adult leaf area index [LAI] and seed rain, newly emerged seedling, and recruited seedling densities) were highly heterogeneous and structured at fine spatial scales. Topography, soil moisture, and light, but not litter thickness, also showed significant spatial autocorrelation at fine spatial scales. However, most of the biotic filters (e.g., seed removal, seed germination, and seedling mortality) were not spatially structured, except for clipping by voles, which was significantly autocorrelated at small distance classes. Failure to detect autocorrelation in most of the biotic filters suggests that they may operate at scales different from that studied, e.g., at a coarser scale for seed predation, but at a finer scale for insect folivory. A path analysis revealed that a large proportion of the variance in seedling recruitment was the result of direct links between the demographic variables studied. Another path analysis, accounting for the effects of biotic filters, highlighted the critical role played by seed predators on recruitment. However, restricted seed dispersal was the major factor influencing the spatial distribution of recruits: indeed, spatial concordance between seed rain and recruit density suggested that even when they are heterogeneous at the same spatial scale, environmental filters are not strong enough to disrupt the patterns generated by primary seed dispersal. Hoarding by rodents accounted for a significant proportion of the spatially homogeneous seed removal; it also reduced potential recruitment limitation associated with microsite availability, thereby increasing the proportion of variation in the spatial pattern of recruitment accounted for by primary seed dispersal.