Adenosylcobalamin (AdoCbl) serves as a reservoir for the 5'-deoxyadenosyl radical, which is generated in enzyme by the homolytic cleavage of a Co-C bond and harnessed to initiate radical reactions by abstracting a hydrogen from the substrate. How these enzymes increase the rate of Co-C bond cleavage by an estimated 12 orders of magnitude, whether the 5' -deoxyadenosyl radical exists as a metastable or transient intermediate and how the first steps of the reaction are coupled are key unresolved questions. The Co-C bond breaking and hydrogen abstraction steps were modelled in AdoCbl dependent glutamate mutase with MD simulations, adiabatic mapping and umbrella sampling simulations using a novel empirical valence bond (EVB) potential, which was calibrated to high level ab initio and DFT calculations. This potential was found to compare favourably with the results of QM/MM calculations. Hydrogen bonding with the protein stabilises the dissociated 5' -deoxyadenosyl radical and induce conformational change, guiding the C5' radical centre towards the substrate hydrogen to be abstracted. The heme dioxygenase enzymes Indoleamine 2,3 -dioxygenase (lDO) and tryptophan 2,3-dioxygenase (TDO) catalyse the first step in the metabolism of L-tryptophan (L-Trp) by insertion of both atoms of heme-bound O2 into the substrate. In an attempt to improve understanding of the differences in substrate binding and reactivity between these enzymes, molecular dynamics (MD) simulations, MMIPBSA binding free energy calculations and reaction modelling with hybrid quantum mechanics/molecular mechanics (QMlMM) adiabatic mapping calculations were performed. Starting with crystal structures for a bacterial TDO (XcTDO) and human IDO (hIDO), reactivity and binding of IDO, TDO and the H55A mutant TDO with L-Trp, D-tryptophan (D-Trp) and I-methyl-L-tryptophan (l -Me-L-Trp) were investigated. Differences in experimental KMs were partially rationalised by analysis of substrate-protein interactions and calculated binding free energies. Although the calculated barriers were unable to rank correctly the active systems, they were able to predict whether a particular system was active, slightly active or inactive. Differences in reactivity were related to the varying ability of the systems to position optimally the substrate in relation to the heme-bound 02.