We carried out an extensive survey of wave function and DFT methods to test their accuracy on geometries and dissociation energies of halogen bonds (XB). For that purpose, we built two benchmark sets (XB18 and XB51). Between the DFT methods, it was found that functionals with high exact exchange or long-range corrections were suitable for these dimers, especially M06-2X, ωB97XD, and double hybrids. Dispersion corrections tend to be detrimental, in spite of the fact that XB is considered a noncovalent interaction. Wave function techniques require heavy correlated methods (i.e., CCSD(T)) or parametrized ones (SCS-MP2 or SCS(MI)MP2). Heavy basis sets are needed to obtain high accuracy, such as aVQZ or aVTZ+CP, and ideally a CBS extrapolation. Relativistic ECPs are also important, even for the bromine based dimers. In addition, we explored some XB with new theoretical tools, the NCI ("Non-Covalent Interactions") method and the NOFF ("Natural Orbital Fukui Functions").