The demise of coppicing in UK ancient woodlands, combined with the planting of non-native, fast-growing conifers in the twentieth century, heightens the potential recharge value of ground flora seed banks. Soil cores from adjoining semi-natural and conifer-containing stands in four lowland ancient woods in central England were removed to establish seed bank species richness. During a fourteen-month germination trial soil from two depths yielded 6554 seedlings from 81 species, ten of which showed a strong affinity for ancient woodland conditions. Juncus effusus accounted for 80% of emergent seeds whilst 23 other species, including Lysimachia nummularia and Potentilla sterilis, were represented by only one individual. Species richness is described by a model that explains 40% of observed variance (P < 0.00001). The model has three significant variables: species richness increases as soil pH rises, and decreases with both depth and increasing time since the most recent planting/disturbance event. No difference was found in the density of seeds from species common to paired semi-natural and conifer-containing stands that were separated only by a woodland ride, suggesting prior management and environmental conditions have a greater influence on seed banks than current stand type. Sørensen similarity index values revealed poor congruence between above-ground vegetation and species in the seed bank. Taking pH measurements in conifer stands identified as younger in terms of planting/disturbance may help locate areas where greater numbers of species (including woodland specialists) are located. Caution is required, however, as these seed banks may also contain non-target, competitive species that may swamp the regeneration of woodland specialists.