ONIOM is a flexible hybrid scheme that can combine the most suitable computational methods for a given system without previous parameterization. The reason for its flexibility is that all calculations are performed on complete molecular systems, and the total energy is obtained from an extrapolation scheme. Most commonly used is the combination of a quantum mechanics and a molecular mechanics method (ONIOM QM:MM), and we describe applications of this method to several enzymatic systems, e.g., glutathione peroxidase and methylmalonyl-CoA mutase. The role of the protein is highlighted by comparing models with and without explicit inclusion of the protein matrix. We also outline future directions for the application of ONIOM to enzymes. One of the major deficiencies of QM/MM models in general, including ONIOM QM:MM, is the poor description of electrostatic interactions between the QM and the MM region. An attractive alternative to QM:MM is to take advantage of the multi-layer capability of ONIOM and design three-layer QM:QM’:MM models. In this scheme QM’ is a relatively fast molecular orbital method that can describe charge transfer and mutual polarization between the reactive region and the protein surroundings