A set of 92 density functionals was employed to accurately characterize thiol-disulfide exchange. The properties we have benchmarked throughout the study include the geometry of a 15 atoms model system, the potential energy surface, the activation barrier, and the energy of reaction for thiol-disulfide exchange. Reference energies were determined at the CCSD(T)/CBS//MP2/aug-cc-pVDZ level of theory, and reference geometries were calculated at the MP2/aug-cc-pVTZ level. M11-L, M06-2X, M06-HF, N12-SX, PBE1PBE, PBEh1PBE, and OHSE2PBE described better the geometry of the model system, with average deviations of 0.06 Å in bond lengths (0.06 Å in bond-breaking lengths) and 1.9° in bond angles. On the other hand, the potential energy surface and its gradient were more accurately described by the hybrid density functional BHandH, closely followed by mPW1N, mPW1K, and mPWB1K. The barrier height and energy of reaction were better reproduced by the BMK and M06-2X functionals (deviations of 0.17 and 0.07 kcal·mol(-1), respectively) for a set of 10 Pople's basis sets. MN12-SX and M11-L showed very good results for the widely used 6-311++G(2d,2p) basis set, with deviations of 0.02 and 0.05 kcal·mol(-1), respectively. We studied the effect of the split-valence, diffuse, and polarized functions in the activation barrier of thiol-disulfide exchange, for a set of 10 Pople's basis sets. While increasing the splitting and polarization may increase the activation barrier in approximately 1 kcal·mol(-1), diffuse functions generally contribute to decreasing it no more than 0.10 kcal·mol(-1). In general, 13 functionals provided energies within 1 kcal·mol(-1) of the reference value. The BB1K density functional is one of the best density functionals to characterize thiol-disulfide exchange reactions; however, several density functionals with modified Perdew-Wang exchange and about 40% Hartree-Fock exchange, such as mPW1K, mPW1N, and mPWB1K, show a good performance, too.