Theoretical investigations of LnX(3) systems (Ln = La, Gd, Lu and X = F, Cl, Br, I) have been carried out by various density functional methods, including nonlocal gradient corrections and self-consistent hybrid density functional/Hartree-Fock approaches. The relativistic effects were taken into account either by a relativistic effective core potential (RECP) or within a frozen core approximation and a quasi-relativistic approach for the valence electrons, either scalar or including the spin-orbit contribution. Geometry optimizations and harmonic frequencies calculations were carried out, as well as computation of the atomization energies. All data were found to be in very good agreement with experimental data, being at least of the same quality as other RECP-based post-Hartree-Fock calculations. The conformation was found to be pyramidal for the lighter lanthanides and halogens and planar for GdBr3, GdI3, LuCl3, LuBr3, and LuI3. Special attention was also devoted to the description of the lanthanide-halogen bond, depending on the hardness of these atoms. The bonding was examined in terms of contributions of the lanthanide atomic orbitals to the molecular orbitals of the LnX(3) species. Charges were calculated through the natural population analysis procedure, and some investigations have been carried out using the natural resonance theory, as implemented in the framework of the natural bond orbital approach. An energetic analysis based on the transition state method of Ziegler et al. was also performed and gave the energetic contributions (i.e., steric, electrostatic, and orbital) to the bonding. All these analyses point to a highly ionic interaction, especially for the lighter halogens and lanthanides, even if some non-negligible ligand-to-metal charge transfer occurs with the more polarizable bromine and iodine. Nevertheless, the stabilization brought by this covalent charactacter is weak compared to the stabilization due to electrostatic interactions.