Soil seed banks that persist after a fire are important in fire-prone habitats as they minimise the risk of decline or local extinction in plants, should the fire-free interval be less than the primary juvenile periods of the species. In two common woody plant genera (Acacia and Grevillea) in southeastern Australia, we examined the size and location of the residual seed bank after fire across areas of varying seedling densities at three locations in comparison to the distribution of seeds in the soil at an unburnt site. We found viable dormant seeds remaining in the soil after fire (evidence of residual soil seed bank). A significantly lower proportion of seeds remained in the top 5 cm of soil than at 5-10 cm or 10-15 cm soil depths, independent of seedling density or plant genus. This was due to greater germination, and possibly some seed mortality, near the soil surface. Reduced germination below 5 cm was probably due to the reduced efficacy of the fire cues that break seed dormancy, a declining ability of seeds to emerge successfully from such depths, and the lower abundance of seeds in the soil at such depths. The magnitude of the residual seed bank was similar across 0-5, 5-10 and 10-15 cm soil depths in Acacia suaveolens. For two Grevillea species, most residual seeds were at 0-5 and 5-10 cm. The residual soil seed bank in the top 10 cm of soil after fire varied across sites with estimates of 0, 19 and 27% in G. speciosa and 23, 35, and 55% in A. suaveolens. At two sites, both species had similar residual seed bank sizes, while at a third, there were large differences between the species (0-55%). The observed patterns imply that the fire-related cues that break seed dormancy generally declined with soil depth. For Acacia, seed dormancy is broken by heat shock, a fire-cue that declines with soil depth. Some 250 species (approx 15% of the fire-prone flora) in the region are thought to have dormancy broken by heat shock. For Grevillea, where seed dormancy is broken by the interaction of smoke and heat shock, at two sites, we suggest three possibilities: (i) the smoke cue declined with soil depth; (ii) both heat and smoke are obligatory for breaking seed dormancy; or (iii) the cues may be independent and additive and below the zone of soil heating, only a proportion of available seeds had dormancy broken by smoke alone. At a third site (no residual seed bank detected) the smoke cue was predicted not to have declined with soil depth. Up to 900 species (just under half the fire-prone flora) in the study region are thought to have seed dormancy broken by the interaction of heat and smoke during the passage of a fire. © Springer 2006. Abstract reproduced with permission.