The forces responsible for the observed geometries of the YX3 (Y = N or P; X = H, F, or Cl) molecules were studied through ab initio computations at the HF-SCF/6-31G*: level. The calculated molecular orbitals were grouped as contributing primarily to (a) the covalent bonds, (b) the terminal atom nonbonding electrons (for X = F or Cl), and (c) the central atom nonbonding electrons. This grouping was accomplished through 3-D plotting and an atomic population analysis of the molecular orbitals. The molecules were then moved through a X-Y-X angular range from 90 degrees to 119 degrees, in four or five degree increments. Single-point calculations were done at each increment, so as to quantify the energy changes in the molecular orbital groups as a function of geometry. These calculations show that the nonbonding electrons are much more sensitive to geometry change than are the bending orbitals, particularly in the trihalide compounds. The molecular orbitals representing the nonbonding electrons on the terminal atoms (both valence and core electrons) contribute to the spreading forces, as they favor a wider X-Y-X angle. The contracting forces, which favor a smaller X-Y-X angle, consist of the orbitals comprising the nonbonding electrons on the central atom (again, both valence and core electrons). The observed geometry is seen as the balance point between these two sets of forces. A simple interaction-distance model of spreading and contracting forces supports this hypothesis. Highly linear trends are obtained for both the nitrogen trihalides (R-2 = 0.981) and phosphorus trihalides (R-2 = 0.992) when the opposing forces are plotted against each other. These results suggest that a revision of the popular conceptual models (hybridization and VSEPR) of molecular geometry might be appropriate.