Stem cell and more differentiated neural cell transplantation strategies are an intriguing approach for neural repair to augment rehabilitation interventions after stroke. In the cortex, exogenous cells could create, augment, or extend in time endogenous peri-infarct and remote molecular signals, such as those for neurogenesis, cell differentiation, axonal and dendritic sprouting, network connectivity, and long-term potentiation, as well as deliver engineered genes and provide replacement cells in a network. If demyelinated axons exist in the periphery of an infarct, they could be targets for remyelination to reestablish conductivity. Much is unknown, however, about the mechanisms by which pluripotent embryonic and multipotent neural stem cells serve as agents of therapeutic plasticity. The robustness of their effects on neuromodulation, reorganization, regeneration, and behavioral recovery is a work in progress. Invasive interventions may have adverse effects not appreciated in preclinical testing. These should initially be offered only to patients with specific profound impairments after it is clinically certain that major disabilities will not improve. If a cellular strategy is very safe, it may be offered to subjects with moderate impairments when they are no longer likely to make further functional gains. Clinical trial designs are suggested that take into account the optimal timing after stroke and specific targets for cellular therapies to foster repair, remapping, and modulation of neural circuits. Cell-mediated rehabilitation would then use task-specific therapies in an optimal dose to maximize training-induced reorganization and learning and, most important, reduce unwanted disability.