A seed size-seed number trade-off exists because smaller seeds are produced in greater number but have a lower probability of establishment. This reduced establishment success of smaller-seeded species may be determined by biophysical constraints imposed by scaling rules. Root and shoot diameter, root growth extension rate (RGER) and shoot length at death for dark-grown seedlings are predicted to scale with the cube root of seed embryo and endosperm mass (m). We confirmed this expectation for ten neotropical gap-dependent tree species with an embryo and endosperm dry mass>1 mg. However, for nine smaller seeded species (m<1 mg) with photoblastic germination, root and shoot diameters were larger than expected, and consequently, RGER was slower than expected. The maximum shoot thrust of seedlings from seeds with masses>or=1 mg was comparable to the estimated force required to displace overlying litter, supporting the hypothesis that photoblastic behaviour only occurs in seeds with insufficient shoot thrust to displace overlying leaves. Using the model soil water, energy and transpiration to predict soil drying in small and large gaps, we showed that: (1) gaps that receive a significant amount of direct sunlight will dry more quickly than small gaps that do not, (2) compared to the wet-season, soil that is already dry at depth (i.e. the dry-season) will dry faster after rainfall (this drying would most likely kill seedlings from small seeds) and (3) even during the wet-season, dry periods of a few days in large gaps can kill shallow-rooted seedlings. We conclude that the smaller the seed, the more vulnerable its seedling would be to both covering by litter and soil drying because it can only emerge from shallow depths and has a slow RGER. Consequently, we suggest that these allometrically related factors contribute to the reduced establishment success of smaller-seeded species that underpins the seed size-seed number trade-off.