The mature oak (Quercus liaotungensis Koidz) forests in the Dongling mountains of northern China have become degraded in recent years because regeneration has been limited. To determine whether or not seedling establishment of the oak is seed limited, microsite limited, or predator limited and to determine whether seedling establishment is affected by ground cover, we conducted field experiments during a mast year and investigated the fate of seeds and the soil seed bank dynamics of the oak. A large acorn crop (128.8 acorns/m2) was observed in the study period, and the peak density of acorns on the forest floor reached 46.5 acorns/m2, suggesting that tree recruitment was not seed limited. Acorns in the soil seed bank were mainly lost through decay (principally after fungal attack), consumption in situ, and removal by animals. Predation (including consumption in situ and removal) accounted for 86.4% of acorn loss and was therefore likely to have been the most important factor influencing seed dynamics. More than 70% of acorns were found to have germinated, but no established seedling was observed on the forest floor. Using cages to exclude predators, it was estimated that 87% of acorns germinated and 49% became established as seedlings, indicating that the acorns on the forest floor could emerge and grow in the absence of predators. We conclude that the regeneration of the tree population is limited by predators rather than by the availability of microsites. The presence of ground cover increased the germination rate and increased the chance of seed survival in the early stage of the experiment, but at the end of the investigation, no established seedling was found in the quadrats both with and without ground cover, possibly because of high density of animal predators. On the basis of these results, we suggest that selective tree felling will increase the coverage of the herbaceous layer, which can further decrease the population density of the rodents, and thereby improve the regeneration of oak trees.