A block-localized wave function method was used to examine the stereoelectronic effects on the origin of the structural difference between trisilylamine and trimethylamine. The pyramidal geometry of trimethylamine along with its high basicity is consistent with the traditional VSEPR (valence shell electron-pair repulsion) model for σ bonding. On the other hand, in trisilylamine, the silicon d orbitals make modest contribution to the electronic delocalization, although the key factor in charge delocalization is still n(N)→σ(SiH)* negative hyperconjugation. Interestingly, the gain in p(π)→d(π) bonding stabilization is offset by a weaker negative hyperconjugation effect in trisilylamine, resulting in an overall smaller delocalization energy (-18.5 kcal/tool) than that in trimethylamine (-23.9 kcal/mol), which contains little p(π)→d(π) bonding character. Significantly, because of the relatively low electronegativity of silicon, the N-Si bond is much more polar than the N-C bond. Weinhold's natural population analyses of the BLW and HF wave functions for these compounds reveal that the origin of the planar geometry of trisilylamine is due to the polar σ-effect that yields significant long-range electrostatic repulsion between the silyl groups. In addition, it was found that only the most electronegative substituents such as F and OH can result in a pyramidal geometry at the nitrogen center for silylamines. This is in good accord with the recent X-ray structure of a pyramidal silylamine, N(CH3)(OCH3)(SiH3).