We present a conceptual framework for choosing native plant material to be used in restoration projects on the basis of ecological genetics. We evaluate both the likelihood of rapid establishment of plants and the probability of long-term persistence of restored or later successional communities. In addition, we consider the possible harmful effects of restoration projects on nearby ecosystems and their native resident populations. Two attributes of the site to be restored play an important role in determining which genetic source will be most appropriate: (1) degree of disturbance and (2) size of the disturbance. Local plants or plants from environments that “match” the habitat to be restored are best suited to restore sites where degree of disturbance has been low. Hybrids or “mixtures” of genotypes from different sources may provide the best strategy for restoring highly disturbed sites to which local plants are not adapted. Cultivars that have been modified by intentional or inadvertent selection have serious drawbacks. Nevertheless, cultivars may be appropriate when the goal is rapid recovery of small sites that are highly disturbed.