Atomistic simulations are a powerful complement to experimental probes for understanding the nanoscale processes associated with the effects of hydrogen (H) on plasticity and fracture that are the underlying causes of hydrogen embrittlement (HE). Current experimental techniques provide quantitative measures of the macroscopic effects of H on plastic flow and fracture but are unable to determine the nanoscale distribution of H atoms in equilibrium nor, more importantly, as a function of time. Conversely, atomistic simulations can provide information on the nanoscale distribution of H around important lattice defects (vacancies, dislocations, grain boundaries, cracks) and probe the mechanical behavior of these defects in the presence and absence of H. Thus, in principle, atomistic simulations can test fundamental theories and conjectures that arise in attempting to rationalize experimental features of HE. However, atomistic simulations have a range of limitations that must be well-recognized. Accurate ab initio simulations are limited to small numbers of atoms and cannot capture necessary time evolution. Molecular simulations using semi-empirical interatomic potentials can handle more atoms and longer time scales, but are limited by accuracy of the potentials and time scales that remain far smaller than experimental time scales. The value of atomistic simulations thus lies primarily in creating targeted simulations to assess the energetics of specific configurations or specific mechanisms of deformation or fracture, along with theoretical models to estimate realistic time scales that remain inaccessible in simulations. Because of their limitations, atomistic simulations may not be definitive, but they nonetheless provide considerable insight by supporting or contradicting conjectures and concepts proposed to rationalize experiments. Here, the above issues are discussed in more detail and several examples, mainly from the work of the current authors, and including previously-unpublished studies on the effects of H on the bowout of the edge dislocations in alpha-Iron and predictions of solute-drag by H in nickel, serve to demonstrate how atomistic simulations can be used to reveal important features of the behavior of H in metals.