By intensive consuming on soil seed bank, the post-dispersal seed predation has prominent impacts on population regeneration and the dynamics of plant community. The spatial variation of seed predation intensity has been generally linked with density and distance to the parent trees in modeling, while the impacts of environmental variations (e.g. topography) have received much less attention, the related contributions and mechanisms are still open. We buried seeds in the field, mimicing the post-dispersal condition, and monitored the seed predation by rodents, in order to test the effects of species preference, topography, time and depth of burial on the probability of predation. In a one year experiment, the cumulated average seed predation is 20%, range from 0 to 48.25% among eight deciduous tree species, the final probability of seed predation at hilltop is about 3 times as those at other topographic positions; the predation probability occurred to the seeds buried in plant litter layer were twice as those buried in soil. The interactions between the factors were analyzed, and a logistic regression model was fitted to estimate their contribution to the variation of seed predation probability. In result, about 45% of variation can be explained by the factors considered. In summary, species preference is a primary factor for the variation of predation, followed by topography, time and burial depth in contributions. Rodents significantly favor the species with larger seeds; their foraging activity seems to be much more intensive or frequent at ridge than other slope positions. Seed predation becomes more intensive since March, and turns down after July. The depth of burial has prominent effect on the efficiency of seed search by rodents.