Although seed harvested from remnant, wildland perennial‐grass populations can be used for restoration in humid and subhumid temperate regions, seed harvested in semiarid and arid environments is often of low quality and highly variable in quantity. In addition, ongoing harvest of indigenous populations can be unsustainable, especially for those that are small. In such environments, dependable and repeatable broad‐scale restoration of degraded grasslands requires sufficient and consistent supplies of reliable, cost‐effective seed sources that can only result from intensively managed cultivated stands. But does the harvest of intensively managed seed‐production fields inadvertently compromise genetic diversity, thereby adversely affecting the restoration outcome? That is, are seed‐production systems a part of the solution for restoration, or do they create new unintended management issues? This article discusses the potential impacts of cultivated seed‐production systems and recurrent artificial selection for specific traits on genetic integrity and performance of native‐species perennial‐grass populations. Although genetic shift resulting from cultivated perennial‐grass seed production may be inevitable, genetic shifts that change phenological expression may be limited in genotypes that exhibit high seed retention. Artificial selection can improve plant material performance on the often‐harsh conditions of restoration sites, but sufficiently high‐effective population sizes (Ne) must be maintained to conserve genetic diversity, thereby precluding the inbreeding depression that can compromise plant performance. Potentially useful traits of native perennial‐grass species that respond to artificial selection include seed production, seed retention, seedling establishment, competitive ability against weeds, and herbicide tolerance. Potential trade‐offs between traits should also be considered to avoid undesirable inadvertent responses to selection.