The mechanism for H(2) cleavage in NiFe-hydrogenase has been reinvestigated with large models using both hybrid DFT by itself, or in a QM/MM scheme following the ONIOM approach. Heterolytic cleavage, with one hydrogen ending up as a bridging hydride and one as a proton on a cysteine ligand, was found to have a barrier slightly too high to be compatible with measured catalytic turnover rates. Alternative mechanisms were therefore investigated. In the finally suggested mechanism, heterolytic cleavage is used only as an initial step to generate a complex with nickel in oxidation state Ni(I). In the following cycles, H(2) is instead cleaved on nickel using an oxidative addition mechanism with a lower barrier. It was found that the ONIOM results for the reaction mechanism in NiFe-hydrogenase needed to be corrected by large model DFT results to be more reliable. This was mainly an effect of overestimation of polarization effects of the QM region by the MM region due to the particular treatment of the electrostatic interactions and the use of a standard (nonpolarizable) force field.