Nonempirical electronic structure theory has been applied to several charge-transfer complexes, which involve ammonia and trimethylamine as electron donors, and molecular fluorine, chlorine, and ClF as electron acceptors. The self-consistent field calculations employed both minimum and double- basis sets of contrcted Gaussian functions. For NH3-F2 and NH3-ClF, the importance of d functions on the N, F, and Cl atoms was investigated. In several cases the minimum basis results do not appear reliable. With the geometries of the donor and acceptor molecules fixed from experiment, the equilibrium geometries of the charge-transfer complexes were predicted. N-X (X = nearest halogen atom) distances are 3.08 (NH3-F2), 2.93 (NH3-Cl2), and 2.65 (NH3-ClF), while the predicted binding energies are 0.6 kcal (NH3-F2), 2.4 kcal (NH3-Cl2), and 7.7 kcal (NH3-ClF). NH3-FCl is predicted to be bound by less than 0.1 kcal/mol. The most intriguing prediction is that the binding energies of the ammonia complexes are greater than those of the corresponding trimethylamine complexes. Although this prediction is in distinct disagreement with accepted chemical intuition, it is consistent with Mulliken populations, which suggest a significantly greater negative charge on the ammonia N atom than that for trimethylamine. Further, the dipole moment of NH3 is significantly larger than that of N(CH3)3. 1975, American Chemical Society. All rights reserved.